From ebcc7fc74cbea7fb6e82b71d181894bcac7ec270 Mon Sep 17 00:00:00 2001 From: Amrit Krishnan Date: Sun, 25 Oct 2026 13:12:57 -0400 Subject: [PATCH 01/10] Upgrade vllm to 0.32.0 - Lock vllm 0.32.0 (torch 2.13.0 cu130, transformers 5.19.0, flashinfer 0.7.0.post1). vllm 0.32 needs Python 3.11+, so the vllm extra is skipped on 3.10, and torchcodec 0.14+. - Move the vllm image to the CUDA 13.0.3 base to match torch's CUDA. - Fix vec-inf metrics for vllm V1 metric names (kv_cache_usage_perc, num_preemptions, prefix cache counters), aggregate across engines and finish reasons, and read SGLang metrics; launch SGLang with --enable-metrics so /metrics exists. - Map -O to --optimization-level, add missing vllm short aliases, and map SGLang --ep-size to --expert-parallel-size. - Do not treat warning-level log lines (e.g. Ray's pubsub warning) as a failed server. --- README.md | 4 +- docs/index.md | 2 +- pyproject.toml | 9 +- tests/vec_inf/cli/test_cli.py | 2 +- tests/vec_inf/cli/test_helper.py | 64 +- tests/vec_inf/client/test_engine_selection.py | 15 +- tests/vec_inf/client/test_helper.py | 88 +- tests/vec_inf/client/test_utils.py | 32 + uv.lock | 4009 +++++++++++++---- vec_inf/cli/_helper.py | 44 +- vec_inf/client/_client_vars.py | 83 +- vec_inf/client/_helper.py | 129 +- vec_inf/client/_slurm_script_generator.py | 5 +- vec_inf/client/_slurm_templates.py | 5 + vec_inf/client/_utils.py | 17 + vllm.Dockerfile | 9 +- 16 files changed, 3444 insertions(+), 1073 deletions(-) diff --git a/README.md b/README.md index bcd67049..560d76ef 100644 --- a/README.md +++ b/README.md @@ -7,11 +7,11 @@ [![code checks](https://github.com/VectorInstitute/vector-inference/actions/workflows/code_checks.yml/badge.svg)](https://github.com/VectorInstitute/vector-inference/actions/workflows/code_checks.yml) [![docs](https://github.com/VectorInstitute/vector-inference/actions/workflows/docs.yml/badge.svg)](https://github.com/VectorInstitute/vector-inference/actions/workflows/docs.yml) [![codecov](https://codecov.io/github/VectorInstitute/vector-inference/branch/main/graph/badge.svg?token=NI88QSIGAC)](https://app.codecov.io/github/VectorInstitute/vector-inference/tree/main) -[![vLLM](https://img.shields.io/badge/vLLM-0.19.0-blue)](https://docs.vllm.ai/en/v0.19.0/) +[![vLLM](https://img.shields.io/badge/vLLM-0.32.0-blue)](https://docs.vllm.ai/en/v0.32.0/) [![SGLang](https://img.shields.io/badge/SGLang-0.5.9-blue)](https://docs.sglang.io/index.html) ![GitHub License](https://img.shields.io/github/license/VectorInstitute/vector-inference) -This repository provides an easy-to-use solution to run inference servers on [Slurm](https://slurm.schedmd.com/overview.html)-managed computing clusters using open-source inference engines ([vLLM](https://docs.vllm.ai/en/v0.19.0/), [SGLang](https://docs.sglang.io/index.html)). **This package runs natively on the Vector Institute cluster environments**. To adapt to other environments, follow the instructions in [Installation](#installation). +This repository provides an easy-to-use solution to run inference servers on [Slurm](https://slurm.schedmd.com/overview.html)-managed computing clusters using open-source inference engines ([vLLM](https://docs.vllm.ai/en/v0.32.0/), [SGLang](https://docs.sglang.io/index.html)). **This package runs natively on the Vector Institute cluster environments**. To adapt to other environments, follow the instructions in [Installation](#installation). ## Installation If you are using the Vector cluster environment, and you don't need any customization to the inference server environment, run the following to install package: diff --git a/docs/index.md b/docs/index.md index 06626b00..e141f6c4 100644 --- a/docs/index.md +++ b/docs/index.md @@ -1,6 +1,6 @@ # Vector Inference: Easy inference on Slurm clusters -This repository provides an easy-to-use solution to run inference servers on [Slurm](https://slurm.schedmd.com/overview.html)-managed computing clusters using open-source inference engines ([vLLM](https://docs.vllm.ai/en/v0.15.0/), [SGLang](https://docs.sglang.io/index.html)). **This package runs natively on the Vector Institute cluster environments**. To adapt to other environments, follow the instructions in [Installation](#installation). +This repository provides an easy-to-use solution to run inference servers on [Slurm](https://slurm.schedmd.com/overview.html)-managed computing clusters using open-source inference engines ([vLLM](https://docs.vllm.ai/en/v0.32.0/), [SGLang](https://docs.sglang.io/index.html)). **This package runs natively on the Vector Institute cluster environments**. To adapt to other environments, follow the instructions in [Installation](#installation). ## Installation diff --git a/pyproject.toml b/pyproject.toml index 766294e7..547c4a26 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -50,12 +50,13 @@ inference = [ [project.optional-dependencies] # vLLM inference backend (conflicts with sglang due to dependency version conflicts) # Install with: uv sync --extra vllm --group inference +# vLLM 0.32 requires Python 3.11 or newer, so the extra skips vllm on Python 3.10 +# (the vec-inf client itself still supports 3.10). vLLM 0.32 needs torchcodec +# 0.14 or newer for video decoding. vllm = [ - "vllm>=0.11.2", + "vllm[audio,bench]==0.32.0; python_version >= '3.11'", "ray[default]>=2.51.0", - "vllm[audio]", - "vllm[bench]", - "torchcodec>=0.9.0,<0.10.0", + "torchcodec>=0.14.0", ] # SGLang inference backend (conflicts with vllm due to dependency version conflicts) # Install with: uv sync --extra sglang --group inference diff --git a/tests/vec_inf/cli/test_cli.py b/tests/vec_inf/cli/test_cli.py index 03cc0a1d..56fcaa17 100644 --- a/tests/vec_inf/cli/test_cli.py +++ b/tests/vec_inf/cli/test_cli.py @@ -306,7 +306,7 @@ def test_metrics_command_server_ready(runner): mock_response.metrics = { "prompt_tokens_per_sec": 100.0, "generation_tokens_per_sec": 500.0, - "gpu_cache_usage": 0.5, + "kv_cache_usage": 0.5, } mock_client.get_metrics.return_value = mock_response diff --git a/tests/vec_inf/cli/test_helper.py b/tests/vec_inf/cli/test_helper.py index a762b08f..af35b136 100644 --- a/tests/vec_inf/cli/test_helper.py +++ b/tests/vec_inf/cli/test_helper.py @@ -259,6 +259,13 @@ def test_output_json_with_error_reasons(self, mock_echo): assert "failed_reason" in json_dict +def _table_rows(table: Table) -> dict[str, str]: + """Return a rich table's rows as {first column: second column}.""" + names = [str(cell) for cell in table.columns[0]._cells] + values = [str(cell) for cell in table.columns[1]._cells] + return dict(zip(names, values)) + + class TestMetricsResponseFormatter: """Test cases for MetricsResponseFormatter.""" @@ -269,9 +276,8 @@ def test_format_metrics_basic(self): "generation_tokens_per_sec": 250.2, "requests_running": 3, "requests_waiting": 1, - "requests_swapped": 0, - "gpu_cache_usage": 0.75, - "cpu_cache_usage": 0.25, + "kv_cache_usage": 0.75, + "preemptions_total": 2, "total_prompt_tokens": 1000, "total_generation_tokens": 2500, "successful_requests_total": 10, @@ -281,20 +287,30 @@ def test_format_metrics_basic(self): formatter.format_metrics() assert isinstance(formatter.table, Table) - assert len(formatter.table.rows) > 0 + rows = _table_rows(formatter.table) + assert rows["Prompt Throughput"] == "100.5 tokens/s" + assert rows["Generation Throughput"] == "250.2 tokens/s" + assert rows["Requests Running"] == "3 reqs" + assert rows["Requests Waiting"] == "1 reqs" + assert rows["KV Cache Usage"] == "75.0%" + assert rows["Preemptions"] == "2" + assert rows["Successful Requests"] == "10 reqs" + # Metrics vLLM V1 no longer exports must not be shown + assert "Requests Swapped" not in rows + assert "GPU Cache Usage" not in rows + assert "CPU Cache Usage" not in rows + # No prefix cache data, no prefix cache row + assert "Prefix Cache Hit Rate" not in rows def test_format_metrics_with_prefix_caching(self): - """Test formatting metrics with prefix caching enabled.""" + """Test the prefix cache hit rate row is shown when reported.""" metrics = { "prompt_tokens_per_sec": 100.0, "generation_tokens_per_sec": 200.0, "requests_running": 2, "requests_waiting": 0, - "requests_swapped": 0, - "gpu_cache_usage": 0.5, - "cpu_cache_usage": 0.1, - "gpu_prefix_cache_hit_rate": 0.8, - "cpu_prefix_cache_hit_rate": 0.6, + "kv_cache_usage": 0.5, + "prefix_cache_hit_rate": 0.8, "total_prompt_tokens": 500, "total_generation_tokens": 1000, "successful_requests_total": 5, @@ -303,9 +319,9 @@ def test_format_metrics_with_prefix_caching(self): formatter = MetricsResponseFormatter(metrics) formatter.format_metrics() - assert formatter.enabled_prefix_caching is True - assert isinstance(formatter.table, Table) - assert len(formatter.table.rows) > 0 + rows = _table_rows(formatter.table) + assert rows["Prefix Cache Hit Rate"] == "80.0%" + assert rows["KV Cache Usage"] == "50.0%" def test_format_metrics_with_latency(self): """Test formatting metrics with latency information.""" @@ -314,10 +330,9 @@ def test_format_metrics_with_latency(self): "generation_tokens_per_sec": 300.0, "requests_running": 1, "requests_waiting": 0, - "requests_swapped": 0, - "gpu_cache_usage": 0.3, - "cpu_cache_usage": 0.1, + "kv_cache_usage": 0.3, "avg_request_latency": 2.5, + "avg_queue_time": 0.125, "total_prompt_tokens": 750, "total_generation_tokens": 1500, "successful_requests_total": 8, @@ -326,8 +341,9 @@ def test_format_metrics_with_latency(self): formatter = MetricsResponseFormatter(metrics) formatter.format_metrics() - assert isinstance(formatter.table, Table) - assert len(formatter.table.rows) > 0 + rows = _table_rows(formatter.table) + assert rows["Avg Request Latency"] == "2.5 s" + assert rows["Avg Queue Time"] == "0.12 s" def test_format_failed_metrics(self): """Test formatting failed metrics with error message.""" @@ -350,18 +366,6 @@ def test_set_metrics_with_dict(self): formatter = MetricsResponseFormatter(metrics_dict) assert formatter.metrics == metrics_dict - def test_check_prefix_caching_enabled(self): - """Test prefix caching detection when enabled.""" - metrics = {"gpu_prefix_cache_hit_rate": 0.8} - formatter = MetricsResponseFormatter(metrics) - assert formatter.enabled_prefix_caching is True - - def test_check_prefix_caching_disabled(self): - """Test prefix caching detection when disabled.""" - metrics = {"prompt_tokens_per_sec": 100.0} - formatter = MetricsResponseFormatter(metrics) - assert formatter.enabled_prefix_caching is False - class TestListCmdDisplay: """Test cases for ListCmdDisplay.""" diff --git a/tests/vec_inf/client/test_engine_selection.py b/tests/vec_inf/client/test_engine_selection.py index d4571086..6064c8af 100644 --- a/tests/vec_inf/client/test_engine_selection.py +++ b/tests/vec_inf/client/test_engine_selection.py @@ -294,8 +294,17 @@ def test_engine_args_compilation_config_vllm_only( launcher = ModelLauncher("test-model", {}) # Should work for vLLM + # vLLM expands -O to --optimization-level , not --compilation-config processed_args = launcher._process_engine_args("-O3", "vllm") - assert processed_args["--compilation-config"] == "3" + assert processed_args == {"--optimization-level": "3"} + assert launcher._process_engine_args("-O=2", "vllm") == { + "--optimization-level": "2" + } + assert launcher._process_engine_args("-O 1", "vllm") == { + "--optimization-level": "1" + } + with pytest.raises(ValueError, match="-O requires an optimization level"): + launcher._process_engine_args("-O", "vllm") # Should raise error for SGLang with pytest.raises(ValueError, match="-O is only supported for vLLM"): @@ -350,7 +359,9 @@ def test_vllm_short_to_long_map(self): assert "-tp" in VLLM_SHORT_TO_LONG_MAP assert VLLM_SHORT_TO_LONG_MAP["-tp"] == "--tensor-parallel-size" assert "-O" in VLLM_SHORT_TO_LONG_MAP - assert VLLM_SHORT_TO_LONG_MAP["-O"] == "--compilation-config" + assert VLLM_SHORT_TO_LONG_MAP["-O"] == "--optimization-level" + assert VLLM_SHORT_TO_LONG_MAP["-cc"] == "--compilation-config" + assert VLLM_SHORT_TO_LONG_MAP["-ep"] == "--enable-expert-parallel" def test_sglang_short_to_long_map(self): """Test SGLang short to long argument mappings.""" diff --git a/tests/vec_inf/client/test_helper.py b/tests/vec_inf/client/test_helper.py index 3a659856..cdea799c 100644 --- a/tests/vec_inf/client/test_helper.py +++ b/tests/vec_inf/client/test_helper.py @@ -132,7 +132,63 @@ def test_process_engine_args_vllm( assert vllm_args["--tensor-parallel-size"] == "4" assert vllm_args["--quantization"] == "awq" - assert vllm_args["--compilation-config"] == "3" + assert vllm_args["--optimization-level"] == "3" + + @patch("vec_inf.client._helper.utils.load_config") + @patch("pathlib.Path.exists") + def test_process_engine_args_vllm_short_aliases( + self, mock_path_exists, mock_load_config, mock_configs + ): + """Test vLLM short aliases, including boolean ones, map to long names.""" + mock_load_config.return_value = mock_configs + mock_path_exists.return_value = True + + launcher = ModelLauncher("test-model", {}) + vllm_args = launcher._process_engine_args( + "-tp=2, -pp=2, -dp=2, -ep, -q=fp8, --max-model-len=4096", "vllm" + ) + + assert vllm_args == { + "--tensor-parallel-size": "2", + "--pipeline-parallel-size": "2", + "--data-parallel-size": "2", + "--enable-expert-parallel": True, + "--quantization": "fp8", + "--max-model-len": "4096", + } + + @patch("vec_inf.client._helper.utils.load_config") + @patch("pathlib.Path.exists") + def test_process_engine_args_value_with_equals( + self, mock_path_exists, mock_load_config, mock_configs + ): + """Test only the first '=' separates the flag from its value.""" + mock_load_config.return_value = mock_configs + mock_path_exists.return_value = True + + launcher = ModelLauncher("test-model", {}) + vllm_args = launcher._process_engine_args( + "--override-generation-config=temperature=0.5", "vllm" + ) + + assert vllm_args == {"--override-generation-config": "temperature=0.5"} + + @patch("vec_inf.client._helper.utils.load_config") + @patch("pathlib.Path.exists") + def test_process_engine_args_sglang_expert_parallel_aliases( + self, mock_path_exists, mock_load_config, mock_configs + ): + """Test --ep and --ep-size both map to SGLang's --expert-parallel-size.""" + mock_load_config.return_value = mock_configs + mock_path_exists.return_value = True + + launcher = ModelLauncher("test-model", {}) + assert launcher._process_engine_args("--ep-size=4", "sglang") == { + "--expert-parallel-size": "4" + } + assert launcher._process_engine_args("--ep=2", "sglang") == { + "--expert-parallel-size": "2" + } @patch("vec_inf.client._helper.utils.load_config") @patch("pathlib.Path.exists") @@ -975,10 +1031,11 @@ def mock_metrics_text(self): vllm:e2e_request_latency_seconds_sum{engine="0",model_name="test-model"} 6.0 vllm:e2e_request_latency_seconds_count{engine="0",model_name="test-model"} 2.0 vllm:request_queue_time_seconds_sum{engine="0",model_name="test-model"} 1.5 +vllm:request_queue_time_seconds_count{engine="0",model_name="test-model"} 2.0 vllm:request_success_total{engine="0",finished_reason="length",model_name="test-model"} 2.0 vllm:num_requests_running{engine="0",model_name="test-model"} 1.0 vllm:num_requests_waiting{engine="0",model_name="test-model"} 0.0 -vllm:gpu_cache_usage_perc{engine="0",model_name="test-model"} 0.123 +vllm:kv_cache_usage_perc{engine="0",model_name="test-model"} 0.123 """ @patch("vec_inf.client._helper.ModelStatusMonitor") @@ -994,26 +1051,6 @@ def test_init(self, mock_status_monitor, mock_status_response): assert collector._prev_prompt_tokens == 0.0 assert collector._prev_generation_tokens == 0.0 assert collector._last_updated is None - assert collector.enabled_prefix_caching is False - - @patch("vec_inf.client._helper.ModelStatusMonitor") - @patch("vec_inf.client._helper.utils.read_slurm_log") - def test_init_check_prefix_caching_enabled( - self, mock_read_slurm_log, mock_status_monitor, mock_status_response - ): - """Test __init__ when enable prefix caching is True.""" - mock_status_monitor.return_value.process_model_status.return_value = ( - mock_status_response - ) - mock_read_slurm_log.return_value = {"enable_prefix_caching": True} - - collector = PerformanceMetricsCollector("12345") - - assert collector.slurm_job_id == "12345" - assert collector._prev_prompt_tokens == 0.0 - assert collector._prev_generation_tokens == 0.0 - assert collector._last_updated is None - assert collector.enabled_prefix_caching is True @patch("vec_inf.client._helper.ModelStatusMonitor") @patch("vec_inf.client._helper.utils.get_base_url") @@ -1096,10 +1133,11 @@ def test_fetch_metrics_second_call( vllm:e2e_request_latency_seconds_sum{engine="0",model_name="test-model"} 6.0 vllm:e2e_request_latency_seconds_count{engine="0",model_name="test-model"} 2.0 vllm:request_queue_time_seconds_sum{engine="0",model_name="test-model"} 1.5 +vllm:request_queue_time_seconds_count{engine="0",model_name="test-model"} 2.0 vllm:request_success_total{engine="0",finished_reason="length",model_name="test-model"} 2.0 vllm:num_requests_running{engine="0",model_name="test-model"} 1.0 vllm:num_requests_waiting{engine="0",model_name="test-model"} 0.0 -vllm:gpu_cache_usage_perc{engine="0",model_name="test-model"} 0.123 +vllm:kv_cache_usage_perc{engine="0",model_name="test-model"} 0.123 """ mock_response.text = updated_metrics metrics = collector.fetch_metrics() @@ -1118,7 +1156,9 @@ def test_fetch_metrics_second_call( assert metrics["successful_requests_total"] == 2.0 assert metrics["requests_running"] == 1.0 assert metrics["requests_waiting"] == 0.0 - assert metrics["gpu_cache_usage"] == 0.123 + assert metrics["kv_cache_usage"] == 0.123 + assert metrics["avg_request_latency"] == pytest.approx(3.0) + assert metrics["avg_queue_time"] == pytest.approx(0.75) @patch("vec_inf.client._helper.ModelStatusMonitor") @patch("vec_inf.client._helper.requests.get") diff --git a/tests/vec_inf/client/test_utils.py b/tests/vec_inf/client/test_utils.py index a7c54e26..ace51ade 100644 --- a/tests/vec_inf/client/test_utils.py +++ b/tests/vec_inf/client/test_utils.py @@ -71,6 +71,38 @@ def test_read_slurm_log_not_found(): (["ERROR: something wrong"], ("FAILED", "ERROR: something wrong")), ([], "LAUNCHING"), (["some other content"], "LAUNCHING"), + # Non-error log levels that mention a failure are not fatal + ( + [ + "[2026-10-25 13:02:09,274 W 1401493 1401493] rpc_client.h:638: " + "Failed to connect to the observability pubsub at address " + "172.17.8.32:12558 within 5 seconds.\n" + ], + "LAUNCHING", + ), + ( + [ + "(EngineCore pid=1) WARNING 10-25 13:01:04 [utils.py:10] " + "Could not find a faster kernel, falling back.\n" + ], + "LAUNCHING", + ), + ( + ["2026-10-25 13:01:48,362\tWARNING worker.py:1 -- Failed to set env\n"], + "LAUNCHING", + ), + # Errors at ERROR level are still fatal + ( + [ + "(EngineCore pid=1) ERROR 10-25 13:01:04 [core.py:1510] " + "EngineCore failed to start.\n" + ], + ( + "FAILED", + "(EngineCore pid=1) ERROR 10-25 13:01:04 [core.py:1510] " + "EngineCore failed to start.", + ), + ), ], ) def test_is_server_running_statuses(log_content, expected): diff --git a/uv.lock b/uv.lock index a2d3cf14..c3a67312 100644 --- a/uv.lock +++ b/uv.lock @@ -2,23 +2,33 @@ version = 1 revision = 5 requires-python = ">=3.10" resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -308,16 +318,16 @@ name = "anthropic" version = "0.93.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "anyio" }, - { name = "distro" }, - { name = "docstring-parser" }, - { name = "httpx" }, - { name = "jiter" }, - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "anyio", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, + { name = "distro", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, + { name = "docstring-parser", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, + { name = "httpx", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, + { name = "jiter", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-sglang') or (python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "sniffio" }, + { name = "sniffio", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/c2/70/2429d6f7c2516db99fb342c3ad89575ab3e0cd31d3d2f6cba5fdf5e9c65b/anthropic-0.93.0.tar.gz", hash = "sha256:fea8376f7d5cdf99d5e8e85a48fe7a7bd8ab307cdfee4b1e8283a18b1c0ce1b5", size = 654155, upload-time = "2026-04-09T18:13:53.522Z" } wheels = [ @@ -330,7 +340,8 @@ version = "4.14.2" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "exceptiongroup", marker = "python_full_version < '3.11'" }, - { name = "idna" }, + { name = "idna", version = "3.15", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "idna", version = "3.20", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11' and python_full_version < '3.13'" }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, ] @@ -344,32 +355,20 @@ name = "apache-tvm-ffi" version = "0.1.9" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and sys_platform != 'darwin'", + "python_full_version < '3.11' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform != 'darwin'", ] dependencies = [ - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/6f/60/1e787a0b5ebf318483235be2a689ee367173983067e441b8379564f667c0/apache_tvm_ffi-0.1.9.tar.gz", hash = "sha256:d2d402587e8906de0a07f4746aa78f3d452c7efe3625d4bb39ac2ad693bce530", size = 2513731, upload-time = "2026-02-27T19:28:06.602Z" } wheels = [ @@ -404,13 +403,12 @@ name = "apache-tvm-ffi" version = "0.1.10" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.12'", - "python_full_version == '3.11.*'", - "python_full_version < '3.11'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", ] dependencies = [ - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/17/b0/5114e30faffe3279a51a5f3b45dd1b7ce09af1246b62447b45a39a374e54/apache_tvm_ffi-0.1.10.tar.gz", hash = "sha256:974c208766c304c780c17c6d405449e862f83b22c7b6b2b8c28b29d55a806ae3", size = 2691605, upload-time = "2026-04-07T19:58:51.767Z" } wheels = [ @@ -440,6 +438,65 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/81/50/900d55d8c3ca5a3fcdcef3a6d999f316d01f9e45e5297c444a2940eff5d2/apache_tvm_ffi-0.1.10-cp314-cp314t-win_amd64.whl", hash = "sha256:25d9130788f9b4563330122503b21e6c0ed37198f1552df36c1561b3704f1b2f", size = 2370990, upload-time = "2026-04-07T19:58:31.855Z" }, ] +[[package]] +name = "apache-tvm-ffi" +version = "0.1.11" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/6d/3d/4b9226cd45aa800a6904603dda9b323d728f3c3869952a673f3483b78b19/apache_tvm_ffi-0.1.11.tar.gz", hash = "sha256:153cd2c5a9717804cb0bcd9b2709f22a1e5f80ed05b5a490faf5949b136eedba", size = 2798354, upload-time = "2026-05-04T17:48:43.852Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/99/07/c17d9608f4c9e4637091210a07e30be0f34b303ecb46bd5a0dd2838a2e0d/apache_tvm_ffi-0.1.11-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:3587eb393096832d356be94b2241c6f13b8f41ff729556ce0dc69a4fc7fed73a", size = 2464945, upload-time = "2026-05-04T17:47:33.715Z" }, + { url = "https://files.pythonhosted.org/packages/e9/91/c9227ef7d42ecd8f48b0ba833d704178325147dfce1d95dca96787eef0bd/apache_tvm_ffi-0.1.11-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:e7527775369a32964e083fed04a3a1ce4134e3e8719a64660ad976be2d0ee58e", size = 2675161, upload-time = "2026-05-04T17:47:36.101Z" }, + { url = "https://files.pythonhosted.org/packages/23/29/87d7931157e47c1a2d5cdb13f306d2c20d064d4e0ba0abd34f6a4b3d8b34/apache_tvm_ffi-0.1.11-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:c5b2b6ae779008ab1586866dc04ab8d5798e5cf6c240df675b78501c6c6f8c95", size = 2789472, upload-time = "2026-05-04T17:47:38.328Z" }, + { url = "https://files.pythonhosted.org/packages/c5/36/0f3cb67ccfb2fa6ab57b14797c4fe524770a5319976cca2a3c185fda2cd1/apache_tvm_ffi-0.1.11-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:66d9a23689070d8c3d3c3a47b3f2624f7b160ab155b6b8283d9b16b1a94e50d1", size = 2585760, upload-time = "2026-05-04T17:47:40.336Z" }, + { url = "https://files.pythonhosted.org/packages/3e/85/934780944c78d3a3c9696f84f1b80d14843a4220a8cbcc90ce0f25e289fb/apache_tvm_ffi-0.1.11-cp310-cp310-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a565f6bf25adf588576578d7e2a272b09afef4084c4b668807045bd9e1ee89a9", size = 2768000, upload-time = "2026-05-04T17:47:42.428Z" }, + { url = "https://files.pythonhosted.org/packages/5c/cc/0f21e0eea204a19cd251a6cc24fe494d07ecdefc11b35be2f8c0e110a1d5/apache_tvm_ffi-0.1.11-cp310-cp310-win_amd64.whl", hash = "sha256:119849c342bd97a9d76ec58eb77dc8dff4ebfbe8b17ea72280c5e5b103277ebe", size = 2407386, upload-time = "2026-05-04T17:47:44.492Z" }, + { url = "https://files.pythonhosted.org/packages/67/df/e573d324e3c7cf77fb526d26d59ec0c365e858f5e09f0742dbc39878a100/apache_tvm_ffi-0.1.11-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:5632f5b4d3af46cb6ccc846120418ad478174e896589ba040bc5a4e7a7356716", size = 2462827, upload-time = "2026-05-04T17:47:46.552Z" }, + { url = "https://files.pythonhosted.org/packages/8f/22/aec1d70baa4bc1e3962a23439f82099f7775992cc5b70a19d4a8ef2a47e4/apache_tvm_ffi-0.1.11-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:2f0d4b165f371d2dee6013e47353d178b01742171fd1092c654cbbc0fa5c6d60", size = 2675459, upload-time = "2026-05-04T17:47:48.63Z" }, + { url = "https://files.pythonhosted.org/packages/9c/b6/84acc663a43ba6e72b4dfd8d923fb5a2d1eb2c867b5b2067c18cb3dc855a/apache_tvm_ffi-0.1.11-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:2cf501753d7693daa73711a27f0f9d9f0f76e9e7d98f2fc2403f423ee7bbfd9b", size = 2789376, upload-time = "2026-05-04T17:47:50.449Z" }, + { url = "https://files.pythonhosted.org/packages/ac/96/e216d5d0f420ccf54775c913b6506755f65bc262511a9948b4d1387bcbc9/apache_tvm_ffi-0.1.11-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:a051c84985be3f9d8a20a16ec4bdba73a7ae01d3fb2f18a2c72bbd7a28aaa155", size = 2584233, upload-time = "2026-05-04T17:47:52.387Z" }, + { url = "https://files.pythonhosted.org/packages/72/9c/af12a5e796a672664f2f18eca989222697b91974a9c9e98d4ec2ecfeeb83/apache_tvm_ffi-0.1.11-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:87f84e7c2393fadac340fd179a631a697effe54d7317b2543e0930452a0a673d", size = 2768221, upload-time = "2026-05-04T17:47:54.057Z" }, + { url = "https://files.pythonhosted.org/packages/3e/f2/7d1c3c13f1cfd479144a41ebcc5b206337b5b02949d0348c739c7fca2079/apache_tvm_ffi-0.1.11-cp311-cp311-win_amd64.whl", hash = "sha256:fd587ecd8ee843bbec467762490c8347af3dfe997608f9841b48a98f5fffac7f", size = 2409048, upload-time = "2026-05-04T17:47:55.824Z" }, + { url = "https://files.pythonhosted.org/packages/05/9d/0f81ca556e5836b3ca64818cdae3f47dc7822bd35d22ddef7a54106d801d/apache_tvm_ffi-0.1.11-cp312-abi3-macosx_11_0_arm64.whl", hash = "sha256:6ae51cc7df415b5f373a9df4baa1165a65608e519bea81e7dd23428f00eeb689", size = 2418793, upload-time = "2026-05-04T17:47:57.879Z" }, + { url = "https://files.pythonhosted.org/packages/2a/a9/f48e5dd4ae1f6f0c5ffac259c0a9531b7d6a7c0a4c45bc2229d55de6adf8/apache_tvm_ffi-0.1.11-cp312-abi3-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:da2c8d07fdc737d1ba75f4de25c29f156905b9dc980f1da90c395b4db525f522", size = 2605176, upload-time = "2026-05-04T17:47:59.676Z" }, + { url = "https://files.pythonhosted.org/packages/36/99/2848df4e8ed5bf51df1d286d1718510584fa61e88adbc9c5b23d71b38f7c/apache_tvm_ffi-0.1.11-cp312-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:78aa1857b04a2ea718317041ab3f01288b3d496e6036eb1b99ebdc9da0fdaef5", size = 2725887, upload-time = "2026-05-04T17:48:01.381Z" }, + { url = "https://files.pythonhosted.org/packages/7d/80/963c991934a4eb0fa0c0178f51963333fe14a96b732009da642b6bf6b42e/apache_tvm_ffi-0.1.11-cp312-abi3-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:a8b845c8dff498fb981c1dda36c954549204191b485a385845e604966594d0b2", size = 2513121, upload-time = "2026-05-04T17:48:03.43Z" }, + { url = "https://files.pythonhosted.org/packages/4d/18/95569107ee83619d61a3bb0d28743a0599f85c5161981e3e098c82c2b185/apache_tvm_ffi-0.1.11-cp312-abi3-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:2843f084cdc94dedacd8b257a395a2b71b8a3dc7fc99711b148bf1d161983128", size = 2697683, upload-time = "2026-05-04T17:48:05.222Z" }, + { url = "https://files.pythonhosted.org/packages/dc/99/f352cf1cce8f6f05584c4adf11de9eca07e6d217229bad6af35fb372926c/apache_tvm_ffi-0.1.11-cp312-abi3-win_amd64.whl", hash = "sha256:bd67e03759d25ff59f4e0ed9c8630a16872afc9dd8792f46ac3c927554015e60", size = 2365545, upload-time = "2026-05-04T17:48:07.295Z" }, + { url = "https://files.pythonhosted.org/packages/27/ae/09242a668eb75ea06282d7cdc3947004cda69040885c340005a23b0aefe3/apache_tvm_ffi-0.1.11-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:f47435e41bf8a2018ef126fad41f18e0c8fe8be4d25fb3ed04b615278b7806d4", size = 2481373, upload-time = "2026-05-04T17:48:09.388Z" }, + { url = "https://files.pythonhosted.org/packages/9d/d1/dc0c26cf68635a1184ba39cccb6cb3cf9675c7030f135f47205e56bdd2b6/apache_tvm_ffi-0.1.11-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:a05b36530d7cd5bb93b1a21a3b81ff060968c20456c4870b1a80d65966d5114f", size = 2639857, upload-time = "2026-05-04T17:48:11.1Z" }, + { url = "https://files.pythonhosted.org/packages/fe/ad/4e3d4c5ec36e2ecadf6e5eb81cde065c69218cf722606b73af0ea6fdab75/apache_tvm_ffi-0.1.11-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:9b158f93bdfc497ead9fce5ffdd4d132708de60970ffc97d890dd62fa39d9fb4", size = 2755683, upload-time = "2026-05-04T17:48:13.016Z" }, + { url = "https://files.pythonhosted.org/packages/51/37/54deceea6bac0e93844bd572a2fae8549e86e6309c732a0acaeb07a88c6b/apache_tvm_ffi-0.1.11-cp314-cp314t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:0f77406e2773ad18109369417b5ccf6aee3c813867dbd5d2d97170bfa7b491f1", size = 2552014, upload-time = "2026-05-04T17:48:14.692Z" }, + { url = "https://files.pythonhosted.org/packages/14/e8/52c9544be5850c7c0e5edce08f2dc9d05c3ecb10b7ae9b3a9313d1b2857e/apache_tvm_ffi-0.1.11-cp314-cp314t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:78f0c9dc69727665de58faebacf6a3f4a1d75a355591e963e1bc691fc9bf5cd5", size = 2730358, upload-time = "2026-05-04T17:48:16.39Z" }, + { url = "https://files.pythonhosted.org/packages/93/b2/afe8a6b8553f51255afdd8063c5d6fd3f4e1978aad424de706440c59fdba/apache_tvm_ffi-0.1.11-cp314-cp314t-win_amd64.whl", hash = "sha256:2f5d417da48dbabbe08933a4d0964b3d2f43d1a4a2c3a6c0092de670c71a8a87", size = 2476516, upload-time = "2026-05-04T17:48:18.022Z" }, +] + [[package]] name = "astor" version = "0.8.1" @@ -565,23 +622,33 @@ name = "ax-platform" version = "1.2.4" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -641,8 +708,7 @@ name = "blake3" version = "1.0.8" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version == '3.11.*'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/75/aa/abcd75e9600987a0bc6cfe9b6b2ff3f0e2cb08c170addc6e76035b5c4cb3/blake3-1.0.8.tar.gz", hash = "sha256:513cc7f0f5a7c035812604c2c852a0c1468311345573de647e310aca4ab165ba", size = 117308, upload-time = "2025-10-14T06:47:48.83Z" } wheels = [ @@ -776,7 +842,7 @@ dependencies = [ { name = "pyro-ppl" }, { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" } }, { name = "threadpoolctl" }, - { name = "torch" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" } }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" } }, ] sdist = { url = "https://files.pythonhosted.org/packages/a0/2e/6f4f5de03baa5e34892383bdd81d9786029d076220396d1f87a6c44ffa07/botorch-0.16.1.tar.gz", hash = "sha256:2a9ec7e43f1acddcf0c170ba084443e128696bc02f0ef9a48cc720f13f71713b", size = 2536271, upload-time = "2025-11-21T16:37:09.42Z" } @@ -794,23 +860,33 @@ name = "botorch" version = "0.17.2" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -823,7 +899,8 @@ dependencies = [ { name = "pyro-ppl" }, { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" } }, { name = "threadpoolctl" }, - { name = "torch" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" } }, ] sdist = { url = "https://files.pythonhosted.org/packages/28/07/0e1b595021f74497677d4b7eba624e806622f0966920fd50e215444580f3/botorch-0.17.2.tar.gz", hash = "sha256:371461f35d882796a87e70ca9a529f31287f9ed984fd9ca3a1d311af58d3fa0c", size = 3747911, upload-time = "2026-03-05T01:08:55.605Z" } @@ -1133,6 +1210,13 @@ wheels = [ name = "click" version = "8.3.2" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] dependencies = [ { name = "colorama", marker = "sys_platform == 'win32'" }, ] @@ -1141,6 +1225,47 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/e4/20/71885d8b97d4f3dde17b1fdb92dbd4908b00541c5a3379787137285f602e/click-8.3.2-py3-none-any.whl", hash = "sha256:1924d2c27c5653561cd2cae4548d1406039cb79b858b747cfea24924bbc1616d", size = 108379, upload-time = "2026-04-03T19:14:43.505Z" }, ] +[[package]] +name = "click" +version = "8.5.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +sdist = { url = "https://files.pythonhosted.org/packages/c7/0e/7fa0ef50764b67090eca4114772a2abf8b6148198475e54c660b97caeee6/click-8.5.0.tar.gz", hash = "sha256:ba0d2089de75ea0310e2dde03160e6ca10009947fb95a182f9b54021bb272e34", size = 382235, upload-time = "2026-08-26T13:33:14.56Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/58/50/6c0d534c5f134586a8e1ba4e330569e32f057e33372ae556463212fb4cd3/click-8.5.0-py3-none-any.whl", hash = "sha256:255bc9599cf7748b4b1a446ccc735421bd08a2ae529a8b88597d3de5664ee360", size = 125251, upload-time = "2026-08-26T13:33:12.928Z" }, +] + [[package]] name = "cloudpickle" version = "3.1.2" @@ -1209,19 +1334,69 @@ wheels = [ name = "compressed-tensors" version = "0.15.0.1" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform != 'darwin'", +] dependencies = [ - { name = "loguru" }, - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "torch" }, - { name = "transformers", version = "5.6.0", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, - { name = "transformers", version = "5.9.0", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-vllm'" }, + { name = "loguru", marker = "python_full_version < '3.11'" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "transformers", version = "5.6.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/41/1b/c3c4a98ec5f2727656336f07a0c35862195c310d8eb0b2fa5b4be6848680/compressed_tensors-0.15.0.1.tar.gz", hash = "sha256:a8e93054e8a5ec49c980b09ed36c4c1249b4a8ee167920a8e461c4da26e78d99", size = 229412, upload-time = "2026-04-10T14:23:54.708Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/a8/52/93833dc1610e017ac5b7dcd59b8304d8ef67d1114c2d124e728a2cbbea12/compressed_tensors-0.15.0.1-py3-none-any.whl", hash = "sha256:e1b1f322e82e475715e242bad46925a304ea8e5c98b5055a15b8eb22fb6bfea9", size = 194260, upload-time = "2026-04-10T14:23:53.098Z" }, ] +[[package]] +name = "compressed-tensors" +version = "0.17.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "loguru", marker = "python_full_version >= '3.11'" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-sglang') or (python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "transformers", version = "5.6.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "transformers", version = "5.19.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/2c/9e/d7f18bd9a0354088abc11a0c1f2c7698f7c49e5a709faedf6a46e388f693/compressed_tensors-0.17.0.tar.gz", hash = "sha256:15c20d06bdbcf35b51fc99fd125e7b9be1e1855567c33b7a46dfac26ad6fb126", size = 257091, upload-time = "2026-06-03T16:49:17.208Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/35/63/6edf0415b072fff0bf8b546074dea3f0f9b148e49b601ac98bdc60a76c68/compressed_tensors-0.17.0-py3-none-any.whl", hash = "sha256:4a1b89b508f7efb8ffb4eee8a6e69e0452d9b080cae130146025c64fbe9fa9aa", size = 211714, upload-time = "2026-06-03T16:49:15.672Z" }, +] + [[package]] name = "contourpy" version = "1.3.2" @@ -1301,23 +1476,33 @@ name = "contourpy" version = "1.3.3" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -1580,7 +1765,8 @@ name = "cuda-bindings" version = "13.2.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "cuda-pathfinder" }, + { name = "cuda-pathfinder", version = "1.5.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "cuda-pathfinder", version = "1.8.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, ] wheels = [ { url = "https://files.pythonhosted.org/packages/1a/fe/7351d7e586a8b4c9f89731bfe4cf0148223e8f9903ff09571f78b3fb0682/cuda_bindings-13.2.0-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:08b395f79cb89ce0cd8effff07c4a1e20101b873c256a1aeb286e8fd7bd0f556", size = 5744254, upload-time = "2026-03-11T00:12:29.798Z" }, @@ -1603,21 +1789,111 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/52/49/4e01cc06447d39476e138d1b1adec8d35c0d04eccd2c8d69befc08cd66e8/cuda_bindings-13.2.0-cp314-cp314t-win_amd64.whl", hash = "sha256:6ccf14e0c1def3b7200100aafff3a9f7e210ecb6e409329e92dcf6cd2c00d5c7", size = 6662637, upload-time = "2026-03-11T00:13:07.881Z" }, ] +[[package]] +name = "cuda-core" +version = "1.2.2" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "cuda-pathfinder", version = "1.8.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "numpy", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/f5/e3/8177d58d0ed609e7213305a693ac9786f2b7fa7f6c3b36dd5ed445a54038/cuda_core-1.2.2-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:3f68e5659cbda6b7d6271a63299778386a8836b99a91061619545163fd15f569", size = 6180837, upload-time = "2026-10-16T13:58:31.772Z" }, + { url = "https://files.pythonhosted.org/packages/c4/9e/9ada2117738871b861cc10c0754cb92171c2190cd122d82aec01c999f551/cuda_core-1.2.2-cp310-cp310-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:2e73547ea81ba0de8af9b48b17da86045735b7d5003251b001553c4eb83c9ebd", size = 6510819, upload-time = "2026-10-16T13:58:34.257Z" }, + { url = "https://files.pythonhosted.org/packages/40/b9/48d936deca5c85936dcd79889d7bb5a8f91c3db40a925dfda872909a4d39/cuda_core-1.2.2-cp310-cp310-win_amd64.whl", hash = "sha256:c8beec95e3db7bca7b45e4c612a233414de658cf2175116e3c1627750398799f", size = 6020662, upload-time = "2026-10-16T13:58:36.522Z" }, + { url = "https://files.pythonhosted.org/packages/05/9a/cd54bb537e33a08f8ae00b8ccf9e4640f716a1c08e3036d6f3fdb619b77d/cuda_core-1.2.2-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:50f2d3e484f50dd7ceaf34e2107aafdf3b7cbcacd73ae38889ca6e477ca3ad5b", size = 6172247, upload-time = "2026-10-16T13:58:38.945Z" }, + { url = "https://files.pythonhosted.org/packages/c4/17/45498f1e74a385eb830f1e1fbee4bda86c30b536287c1c944215d792374a/cuda_core-1.2.2-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:55a0ea9e590e54e696f04f20b037f52dbac3af09bce0039602324cd4151daab9", size = 6508937, upload-time = "2026-10-16T13:58:42.572Z" }, + { url = "https://files.pythonhosted.org/packages/97/81/97d75b23c7d4d63b2a548adbecf35313eb6d7617d046a93aa87d940c6061/cuda_core-1.2.2-cp311-cp311-win_amd64.whl", hash = "sha256:3c31e986bcaecb221ef520020f50ba4650d2a2de51fd1f88cffb1df55176d98a", size = 6028699, upload-time = "2026-10-16T13:58:44.728Z" }, + { url = "https://files.pythonhosted.org/packages/71/17/ecc3da3fe32e1d56e120d9c2316c62aa509cf5660ae970d5a11eb72d3155/cuda_core-1.2.2-cp311-cp311-win_arm64.whl", hash = "sha256:e0f70cde75b11a2de47cebf17b3479442dea02aae9cc43957175dcdfd5680e5c", size = 3030692, upload-time = "2026-10-16T13:58:46.601Z" }, + { url = "https://files.pythonhosted.org/packages/58/f0/437869ea39a6c97468d1141dcee862ccc3e409e8afabd78896e42213c867/cuda_core-1.2.2-cp312-cp312-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:29f0b14a513374e47c6bbd7f1185d735a9a8c3573abf8da4e836805c3d6c7d82", size = 6212012, upload-time = "2026-10-16T13:58:48.968Z" }, + { url = "https://files.pythonhosted.org/packages/f2/1c/8e1f7d42984d6b164521a1dec587f3cf6d1fc585ff929f1420faa07da9d3/cuda_core-1.2.2-cp312-cp312-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0b299eee52edece08b77428cfc22b33b29be1bc69c42878d2e6114031eac17e6", size = 6610599, upload-time = "2026-10-16T13:58:51.223Z" }, + { url = "https://files.pythonhosted.org/packages/4a/7c/22f247c9357c6457281f4f51cceeb023cb47f1071a1d176a07c587e9f849/cuda_core-1.2.2-cp312-cp312-win_amd64.whl", hash = "sha256:d4e6aebf81ade4dd5cf8e2dcf3af25567921bb9439af051f1c2b250f0185d21b", size = 5955406, upload-time = "2026-10-16T13:58:53.333Z" }, + { url = "https://files.pythonhosted.org/packages/67/b3/c1ed3ecb5bc1b11b1ffaeed369ba19fb5c1fec3a4204307baa127b9a888f/cuda_core-1.2.2-cp312-cp312-win_arm64.whl", hash = "sha256:c06a0cdbaba040833cff49f0671d5fe2465d553b839b65967e5f25065b5a8847", size = 2984288, upload-time = "2026-10-16T13:58:55.185Z" }, + { url = "https://files.pythonhosted.org/packages/4b/54/3d7e37e945ef3dd470cce76c1f4a239f26811da15409e4144a9b5f85af73/cuda_core-1.2.2-cp313-cp313-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1094bb311722d3dde862235645244451f39f07f34ce310b894d3f02d0acafbb0", size = 6155990, upload-time = "2026-10-16T13:58:57.093Z" }, + { url = "https://files.pythonhosted.org/packages/24/96/a37f7bccb4ae5057ba135b54186f6d6fc48d219b789f65b597b268fc783b/cuda_core-1.2.2-cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:365baf1e7ac2227f3d847dd83f0cdc75b7d05fa7580f2926e05caca56cea8ea3", size = 6564139, upload-time = "2026-10-16T13:58:59.12Z" }, + { url = "https://files.pythonhosted.org/packages/cc/b4/d0564ed8a96047d820f7ca144906d6f7c9ac4d29d8af80989c2ca4920b8f/cuda_core-1.2.2-cp313-cp313-win_amd64.whl", hash = "sha256:4f346d42c8df86aa75d76f2d41c01efff719cb55beb7b02bbdadf19f9eee3901", size = 5925258, upload-time = "2026-10-16T13:59:01.018Z" }, + { url = "https://files.pythonhosted.org/packages/ad/8f/ad8404e91432028561bd54d290c03b719ca6acaf2ce66e73506dfa39b479/cuda_core-1.2.2-cp313-cp313-win_arm64.whl", hash = "sha256:f43ebc01ed8143d2fc7a2ae5bec1c64f96b35d72f55f728198ded869605bcbfd", size = 2963620, upload-time = "2026-10-16T13:59:02.879Z" }, + { url = "https://files.pythonhosted.org/packages/48/0c/92d7b6a59918f11507f3f0da2279c4b2119727414d5ddef0ab480502a065/cuda_core-1.2.2-cp314-cp314-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:7911da1a70e9d6b115db16f4995c4c00b9bb18457e4e5145b3ff27aa4dc291cd", size = 6234913, upload-time = "2026-10-16T13:59:05.058Z" }, + { url = "https://files.pythonhosted.org/packages/65/d4/5ea4865f80fe6789d89316c2d1270655e6e7da613fec1a3e1ec7b81e52f4/cuda_core-1.2.2-cp314-cp314-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5fd8ac1b7138babe1dc5476650aaab54810292885e5db9e18808ba616f4ccdce", size = 6597619, upload-time = "2026-10-16T13:59:07.246Z" }, + { url = "https://files.pythonhosted.org/packages/61/88/84c8aaf4717b3672b4b58558a226a56e37eb0c59144338bde4b15157a68c/cuda_core-1.2.2-cp314-cp314-win_amd64.whl", hash = "sha256:aba19b094d58931142a890b22b5cdd3ce4568a23b737da42b42cf6c1dcd859f6", size = 6053478, upload-time = "2026-10-16T13:59:09.273Z" }, + { url = "https://files.pythonhosted.org/packages/d0/b3/b8f773f9fab701e800a7a0beaf3cdb2d4f2d035b6ce87ee537131612f533/cuda_core-1.2.2-cp314-cp314-win_arm64.whl", hash = "sha256:9ef12d40b381dfb098b444285ccdcda4492e6c062d120e315b24029c8c09e445", size = 3062069, upload-time = "2026-10-16T13:59:11.784Z" }, + { url = "https://files.pythonhosted.org/packages/66/50/72f972eba3f014523fec0f58a468fecd7e4d013293013075550b78b29edb/cuda_core-1.2.2-cp314-cp314t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:202679860274849e0fd61ecb8390286555ec764f0772db7ccc7768cfc1406d6c", size = 6409423, upload-time = "2026-10-16T13:59:14.148Z" }, + { url = "https://files.pythonhosted.org/packages/c2/03/39a417b4292664a535e8acb9a9009b0e10f31a13d9861db9c2d83f552763/cuda_core-1.2.2-cp314-cp314t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:e833404527908776bc46991484f825dede7df7e9bca544334a292e845c794eb8", size = 6684790, upload-time = "2026-10-16T13:59:16.114Z" }, + { url = "https://files.pythonhosted.org/packages/d8/44/a53db2712b7bbe81afe78c37eb567e9f92d4d5b6b34f6a299278fd8f3c75/cuda_core-1.2.2-cp314-cp314t-win_amd64.whl", hash = "sha256:25dc19a7e3cfb1cb0a602361141a62c2b48771ac649e2e4ea9e16266c2ffb51d", size = 6974148, upload-time = "2026-10-16T13:59:18.191Z" }, + { url = "https://files.pythonhosted.org/packages/a4/93/d65888f880a95901f9d285d2a62d3d219a8661dcce6b5f48a4568393cc7a/cuda_core-1.2.2-cp314-cp314t-win_arm64.whl", hash = "sha256:287a6f19c2fdec64242dcc49605b95bb2b9432374ab33b632374c8dc1bfefe7b", size = 3232596, upload-time = "2026-10-16T13:59:20.386Z" }, + { url = "https://files.pythonhosted.org/packages/6c/22/36720c12a8ff7df3e20cfc26e9abd35270038cfa8f676734e79cbe93b514/cuda_core-1.2.2-cp315-cp315-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:112fd0fe6f190580f05dcf47d8e74aa664cfac9232c47520b6b10a5dd6873c96", size = 6247325, upload-time = "2026-10-16T13:59:22.626Z" }, + { url = "https://files.pythonhosted.org/packages/b0/c4/53e6620f2d9eb7e5b24395d55470963acb1c9ff4eb4f1468a3647ea35fdc/cuda_core-1.2.2-cp315-cp315-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4c093f2b5a952e6a8150d5420ef46241707f8a8d6821a3e6fa23e22019d95e50", size = 6598997, upload-time = "2026-10-16T13:59:24.917Z" }, + { url = "https://files.pythonhosted.org/packages/d8/ab/1127d63cf70ac443ec8573cf470a170d714fe879a87d0caaf128da479b02/cuda_core-1.2.2-cp315-cp315-win_amd64.whl", hash = "sha256:b025faca9b6d4a78d03a3d54f2fefddc087383213665b0b43df379460d6b8e53", size = 6054424, upload-time = "2026-10-16T13:59:27.18Z" }, + { url = "https://files.pythonhosted.org/packages/ac/ce/1affa56f6165edd3891f107c60d174a526b7ffb78e55e99f50156937b721/cuda_core-1.2.2-cp315-cp315-win_arm64.whl", hash = "sha256:08537b22c7f1bc9383956c1e66262b8130d660d7785730e1b1020bdfc3b6ebe5", size = 3062426, upload-time = "2026-10-16T13:59:29.283Z" }, + { url = "https://files.pythonhosted.org/packages/a0/16/0db636be94e3be753b747eb3f6c0e55ee1666f9f488757c9741109000926/cuda_core-1.2.2-cp315-cp315t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:14ddf9fbf7ffc486d4abc9e147a593e2ba8d8004699bea137dffb704a255d4b6", size = 6405060, upload-time = "2026-10-16T13:59:31.419Z" }, + { url = "https://files.pythonhosted.org/packages/5e/11/60cfc4cb9ad1a6e1d9501478ac652769089fb903cffcab1407864bfa7451/cuda_core-1.2.2-cp315-cp315t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:25a823c4c0a9007e46662b9612dfdd5252fe7955c50e04861bc4cbdaa0e7062f", size = 6674158, upload-time = "2026-10-16T13:59:33.63Z" }, + { url = "https://files.pythonhosted.org/packages/62/62/291a0f34095409923ebc4bd1e23f50ce705545e8f7e9c2facb2c97993dcd/cuda_core-1.2.2-cp315-cp315t-win_amd64.whl", hash = "sha256:d1feed1fa543c01cd115d31cbb7d74145d742b1252b2a748c71e37cb2ce194b2", size = 6953658, upload-time = "2026-10-16T13:59:35.853Z" }, + { url = "https://files.pythonhosted.org/packages/96/b5/8ddc268eb9ee32e4f8e2a1a01dfa1518d04e72122c9522810ee85ecbf5b7/cuda_core-1.2.2-cp315-cp315t-win_arm64.whl", hash = "sha256:95a9228d6a82f38838f493eecf205673a181833baa97961a1ee625ada67ea3c8", size = 3226305, upload-time = "2026-10-16T13:59:37.725Z" }, +] + [[package]] name = "cuda-pathfinder" version = "1.5.2" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] wheels = [ { url = "https://files.pythonhosted.org/packages/f2/f9/1b9b60a30fc463c14cdea7a77228131a0ccc89572e8df9cb86c9648271ab/cuda_pathfinder-1.5.2-py3-none-any.whl", hash = "sha256:0c5f160a7756c5b072723cbbd6d861e38917ef956c68150b02f0b6e9271c71fa", size = 49988, upload-time = "2026-04-06T23:01:05.17Z" }, ] +[[package]] +name = "cuda-pathfinder" +version = "1.8.4" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/8e/93/f0904fa7d954ecc4c0db17c8d63d185d442dd59fc6757f1d1f448c59c256/cuda_pathfinder-1.8.4-py3-none-any.whl", hash = "sha256:85a4e708c725d603f576042ef32781a3d1737c51e84c35fe361d2680191af65a", size = 62594, upload-time = "2026-10-22T21:49:19.246Z" }, +] + [[package]] name = "cuda-python" version = "13.2.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "cuda-bindings" }, - { name = "cuda-pathfinder" }, + { name = "cuda-pathfinder", version = "1.5.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "cuda-pathfinder", version = "1.8.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, ] wheels = [ { url = "https://files.pythonhosted.org/packages/4a/da/b4dbe129f941afe1c24a09ba53521b78875626763d96414798a74763282f/cuda_python-13.2.0-py3-none-any.whl", hash = "sha256:2f092b0ec13a860115fa595411889ee939ad203450ea4f91e9461b174ea7b084", size = 8145, upload-time = "2026-03-11T13:55:19.143Z" }, @@ -1627,9 +1903,15 @@ wheels = [ name = "cuda-tile" version = "1.3.0" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] dependencies = [ - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" } }, ] wheels = [ { url = "https://files.pythonhosted.org/packages/46/c8/1687a83d0739151ca410a5716b5f1dfb6f8feb6381c7c695d72264f17e1c/cuda_tile-1.3.0-cp310-cp310-manylinux2014_aarch64.whl", hash = "sha256:c55616c648f06f84808648a521c67f2d7c790574d6b53ddf8c3bfbc995d36d45", size = 245438, upload-time = "2026-04-20T15:51:18.862Z" }, @@ -1646,47 +1928,182 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/2b/77/c13afad1a06824c1c942afd0205e78ff17f0ee06fc1a943f6e2135cf4112/cuda_tile-1.3.0-cp313-cp313-win_amd64.whl", hash = "sha256:efcb93c25563fe23d6aa083c22893fd703122eaf684b0d36874982d28a6dad0b", size = 240925, upload-time = "2026-04-20T15:52:21.283Z" }, ] +[[package]] +name = "cuda-tile" +version = "1.7.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" } }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/8f/66/71ef3a678ee84e1e4abe3052f72f6660e704a35f002c34d81b99320efde1/cuda_tile-1.7.0-cp310-cp310-manylinux2014_aarch64.whl", hash = "sha256:7838f48c59e4e3c6868a1a09ebcca3154ea7e1b7cf36bd85cd86439e366b65ad", size = 422572, upload-time = "2026-10-19T16:23:09.029Z" }, + { url = "https://files.pythonhosted.org/packages/be/3f/1f93c3bb83f9af8371f81973bafe5d24f281cfeec60b7a25014d65361ebe/cuda_tile-1.7.0-cp310-cp310-manylinux2014_x86_64.whl", hash = "sha256:87d70d15eef71200e1dd251162828e10dbf4d4e18d77b0a191dd62520bd48568", size = 426219, upload-time = "2026-10-19T16:19:20.733Z" }, + { url = "https://files.pythonhosted.org/packages/42/8c/1bab88304f0d302d0d62300dcd93f0cf9ec4bf6ce5db211376876edf54dd/cuda_tile-1.7.0-cp310-cp310-win_amd64.whl", hash = "sha256:ee6f9994c3a4b1bc824291fe820c2ecedcd485e31e6624a8548746e2646a8a68", size = 391327, upload-time = "2026-10-19T16:26:45.229Z" }, + { url = "https://files.pythonhosted.org/packages/18/5f/6104c39db3b1f30dfcdca2dfdceb14497c0343900cdfca288e9f2dda82a1/cuda_tile-1.7.0-cp311-cp311-manylinux2014_aarch64.whl", hash = "sha256:c7f2539add95e3f5bf115dc7d33ac7f585ae679c8d7e0443e61274b287be3b99", size = 422278, upload-time = "2026-10-19T16:23:47.538Z" }, + { url = "https://files.pythonhosted.org/packages/f9/93/bb8bb516553a5a220463b71bacabfd0f578ccd3cc36681f34689e294faca/cuda_tile-1.7.0-cp311-cp311-manylinux2014_x86_64.whl", hash = "sha256:8a7b5392486c66c7a5668975740d514aee03fb5c38bd6657445091d5076d0291", size = 426167, upload-time = "2026-10-19T16:22:34.728Z" }, + { url = "https://files.pythonhosted.org/packages/79/69/aee31490e8e7941167ff7992fda3d252f9200519a6e32c7eb1eb9c855a1b/cuda_tile-1.7.0-cp311-cp311-win_amd64.whl", hash = "sha256:78d9b2df869e036087cc60d69770ae35d2d3be85fa8220ec34c024731ca0aed4", size = 390858, upload-time = "2026-10-19T16:27:37.133Z" }, + { url = "https://files.pythonhosted.org/packages/c3/9a/e384f42ce51b889f3124a5d0fe061c2cf8348305f8caca3d43594b7b53b9/cuda_tile-1.7.0-cp311-cp311-win_arm64.whl", hash = "sha256:83c1036c5eb48761a92af0307bf90b7f5f5c7ce9bcf6d5f04aafc444eff49a1c", size = 380435, upload-time = "2026-10-19T16:30:00.609Z" }, + { url = "https://files.pythonhosted.org/packages/c6/d9/e356e8db19c230c67fad43ff7d55be1c3b21d0673d4e45ba9724d86c7b17/cuda_tile-1.7.0-cp312-cp312-manylinux2014_aarch64.whl", hash = "sha256:0b5530e3f9e8be3de903bb726d759ed426674653afa34fe507ff83791ae8050f", size = 422551, upload-time = "2026-10-19T16:25:37.17Z" }, + { url = "https://files.pythonhosted.org/packages/62/a1/071e186cc45859df593b80d292e03581af30755cadebbd180603b0fc6712/cuda_tile-1.7.0-cp312-cp312-manylinux2014_x86_64.whl", hash = "sha256:f886e84a7b54de5e26598eb9a828387a4c87e6ab5b986838cd8a5277a9ffab40", size = 426980, upload-time = "2026-10-19T16:27:12.765Z" }, + { url = "https://files.pythonhosted.org/packages/44/33/6cfcb215db17ecbd03aca9ccd1121da57018dd0c0258eef078300b508a96/cuda_tile-1.7.0-cp312-cp312-win_amd64.whl", hash = "sha256:e72a0f03513cb504d4230099f836219888e2c403086b5ca084717dd5b9c230e2", size = 391816, upload-time = "2026-10-19T16:27:58.832Z" }, + { url = "https://files.pythonhosted.org/packages/d9/68/433a2900c7551d1b2cac35a4d87b04e6d521bd033de00aeccc3e9a5f8f5b/cuda_tile-1.7.0-cp312-cp312-win_arm64.whl", hash = "sha256:fec9546736124a4127a9db55307ccb363f9be38d4248a262b1003c2f3ed4a3f3", size = 380141, upload-time = "2026-10-19T16:30:28.256Z" }, + { url = "https://files.pythonhosted.org/packages/e1/d1/0668037f2eb24c05e1478c0c630a108797803eae2aec49530a3f6dbf9796/cuda_tile-1.7.0-cp313-cp313-manylinux2014_aarch64.whl", hash = "sha256:780e48eec04419fef79476690307415429e42af1232511eba5973c35db17076f", size = 422551, upload-time = "2026-10-19T16:24:57.792Z" }, + { url = "https://files.pythonhosted.org/packages/70/76/486f77b44f3f2407e98fb3b158f91fe0782050a241e0c20798690084d14a/cuda_tile-1.7.0-cp313-cp313-manylinux2014_x86_64.whl", hash = "sha256:a0cdc35796df8caf8fb1850e69d9d7d56e9a8da11e8bcb3042a64f1528f34b4d", size = 426989, upload-time = "2026-10-19T16:21:43.086Z" }, + { url = "https://files.pythonhosted.org/packages/8f/d5/6e78b090f2efd22ba046a4febf12b7a9aee863753a67fab1aa173d02f06d/cuda_tile-1.7.0-cp313-cp313-win_amd64.whl", hash = "sha256:a6c59f85b19925e939b7c8018f3851ce031d7969b5e8d1de5fbf0e43d365d25d", size = 391812, upload-time = "2026-10-19T16:28:25.582Z" }, + { url = "https://files.pythonhosted.org/packages/7b/9f/5499a684f6c29d0fcd27deaa914a062d10b7dac344ba8dad205b69aa365c/cuda_tile-1.7.0-cp313-cp313-win_arm64.whl", hash = "sha256:f75a0c58e95b0cb8026a47c7bd4b11f63434c3366646cef1644da49a18997855", size = 380142, upload-time = "2026-10-19T16:30:49.011Z" }, + { url = "https://files.pythonhosted.org/packages/aa/1c/0d50e18792b571c4c9ecbb5ae61bcb443d2f4de4e411a16c1b5334b3a975/cuda_tile-1.7.0-cp314-cp314-manylinux2014_aarch64.whl", hash = "sha256:f556d5801e523a5b4080a04069d4f05f319db5dfcda232ba46a800fe8e219fa8", size = 422981, upload-time = "2026-10-19T16:20:00.086Z" }, + { url = "https://files.pythonhosted.org/packages/f6/82/557c8bb7a73381cc287106d3f35a39858ea876f54af0805c9a9c3e8f958b/cuda_tile-1.7.0-cp314-cp314-manylinux2014_x86_64.whl", hash = "sha256:8ceec7a4fceb9c4705ac4d98817ca9da1ee7c543dd32465fb74b37d15906f341", size = 427543, upload-time = "2026-10-19T16:24:27.983Z" }, + { url = "https://files.pythonhosted.org/packages/fa/df/78a4a9121dcb819ad37604af397d3fd79d03c92019e96b7aaaae0161ef71/cuda_tile-1.7.0-cp314-cp314-win_amd64.whl", hash = "sha256:c6e61951b7708e21834cda11c5aebe98a7b2d57b609f299f30741c1de74c9630", size = 394551, upload-time = "2026-10-19T16:28:59.057Z" }, + { url = "https://files.pythonhosted.org/packages/5a/06/3496499dcaf075f753259efd12646dc7ddc542938de35822ad7cb2926989/cuda_tile-1.7.0-cp314-cp314-win_arm64.whl", hash = "sha256:2cedf11788c5ab60436a3ee7ed2b04f10463e388fa835417734de4b65faf328b", size = 381421, upload-time = "2026-10-19T16:31:16.262Z" }, + { url = "https://files.pythonhosted.org/packages/6b/73/083c854de98b1a803b85e4c2d003f39c0c45ee32d64e285b31a952aabad2/cuda_tile-1.7.0-cp314-cp314t-manylinux2014_aarch64.whl", hash = "sha256:f8cbcd5a6ec96baf41f46b3ee0836e0a564cc5049324ea94aa6184193c89485f", size = 427763, upload-time = "2026-10-19T16:26:14.184Z" }, + { url = "https://files.pythonhosted.org/packages/2a/0c/10dce345c1fa656668a59ddb85af9e88168998600d57d207947011aee8a3/cuda_tile-1.7.0-cp314-cp314t-manylinux2014_x86_64.whl", hash = "sha256:c72497780587d59f2444e7ae3298aa521425fdda00b0fd347a3dc2986eeac57c", size = 431950, upload-time = "2026-10-19T16:20:52.43Z" }, + { url = "https://files.pythonhosted.org/packages/fd/22/17a40b2d0ec6e31c577d06bd243d0b5d9bfed011430440d905f39220ba10/cuda_tile-1.7.0-cp314-cp314t-win_amd64.whl", hash = "sha256:9bd257ecbe4c7a9fa417d0439d885c0c25e4818a6e34c3a38d1eb6f29ea547dc", size = 405194, upload-time = "2026-10-19T16:29:32.263Z" }, + { url = "https://files.pythonhosted.org/packages/9b/32/bf1d0db1cf58c313925f6a394dc2dcc93988e0a2e572305a7f31a69fdd05/cuda_tile-1.7.0-cp314-cp314t-win_arm64.whl", hash = "sha256:564c26f4cc212770d208f18c7078de947a01eb1b068e07c09d97a86b6f7044c1", size = 388266, upload-time = "2026-10-19T16:31:44.566Z" }, +] + [[package]] name = "cuda-toolkit" version = "13.0.2" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] wheels = [ { url = "https://files.pythonhosted.org/packages/57/b2/453099f5f3b698d7d0eab38916aac44c7f76229f451709e2eb9db6615dcd/cuda_toolkit-13.0.2-py2.py3-none-any.whl", hash = "sha256:b198824cf2f54003f50d64ada3a0f184b42ca0846c1c94192fa269ecd97a66eb", size = 2364, upload-time = "2025-12-19T23:24:07.328Z" }, ] [package.optional-dependencies] cublas = [ - { name = "nvidia-cublas" }, + { name = "nvidia-cublas", version = "13.1.0.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +cudart = [ + { name = "nvidia-cuda-runtime", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +cufft = [ + { name = "nvidia-cufft", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +cufile = [ + { name = "nvidia-cufile", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +cupti = [ + { name = "nvidia-cuda-cupti", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +curand = [ + { name = "nvidia-curand", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +cusolver = [ + { name = "nvidia-cusolver", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +cusparse = [ + { name = "nvidia-cusparse", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +nvjitlink = [ + { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +nvrtc = [ + { name = "nvidia-cuda-nvrtc", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +nvtx = [ + { name = "nvidia-nvtx", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] + +[[package]] +name = "cuda-toolkit" +version = "13.0.3.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/d1/c7/a79086a62c98befcdb8349656c6f114e2db3b8b2422f6e25c97a7f2a9a3c/cuda_toolkit-13.0.3.0-py2.py3-none-any.whl", hash = "sha256:d693caaa261214ddd7dbb60d68e71cbed884e68c2be7509778f3051da0b91c3f", size = 2512, upload-time = "2026-04-14T00:50:08.173Z" }, +] + +[package.optional-dependencies] +cublas = [ + { name = "nvidia-cublas", version = "13.1.1.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cuda-nvrtc", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] cudart = [ - { name = "nvidia-cuda-runtime" }, + { name = "nvidia-cuda-runtime", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] cufft = [ - { name = "nvidia-cufft" }, + { name = "nvidia-cufft", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-nvjitlink", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] cufile = [ - { name = "nvidia-cufile" }, + { name = "nvidia-cufile", marker = "(python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] cupti = [ - { name = "nvidia-cuda-cupti" }, + { name = "nvidia-cuda-cupti", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] curand = [ - { name = "nvidia-curand" }, + { name = "nvidia-curand", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] cusolver = [ - { name = "nvidia-cusolver" }, + { name = "nvidia-cublas", version = "13.1.1.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cusolver", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cusparse", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-nvjitlink", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] cusparse = [ - { name = "nvidia-cusparse" }, + { name = "nvidia-cusparse", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-nvjitlink", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] nvjitlink = [ - { name = "nvidia-nvjitlink" }, + { name = "nvidia-nvjitlink", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] nvrtc = [ - { name = "nvidia-cuda-nvrtc" }, + { name = "nvidia-cuda-nvrtc", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] nvtx = [ - { name = "nvidia-nvtx" }, + { name = "nvidia-nvtx", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and platform_machine == 'AMD64' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] [[package]] @@ -1694,7 +2111,8 @@ name = "cupy-cuda12x" version = "14.0.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "cuda-pathfinder" }, + { name = "cuda-pathfinder", version = "1.5.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "cuda-pathfinder", version = "1.8.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "numpy" }, ] wheels = [ @@ -1715,6 +2133,24 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/e5/a3/80ff83dcad1ac61741714d97fce5a3ef42c201bb40005ec5cc413e34d75f/cupy_cuda12x-14.0.1-cp314-cp314-win_amd64.whl", hash = "sha256:cafe62131caef63b5e90b71b617bb4bf47d7bd9e11cccabea8104db1e01db02e", size = 96822848, upload-time = "2026-02-20T10:23:42.684Z" }, ] +[[package]] +name = "cwevals" +version = "0.0.1b1" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "anyio", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "distro", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "httpx", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and platform_machine == 's390x') or (python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and platform_machine == 's390x') or (python_full_version >= '3.12' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "sniffio", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/91/82/d176b1832e6632c5ebdabbbb943655902f3ff867a18262a0b6cb0e9bf560/cwevals-0.0.1b1.tar.gz", hash = "sha256:2ae997369ff44941925e4db1bd07eb4e13c05de412e615636fd12881d63f8a13", size = 258563, upload-time = "2026-09-28T18:21:23.442Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/b5/96/b3b348954f88915e60e2f1fbe6d09250584e53a20d949e55786f31dbaeb0/cwevals-0.0.1b1-py3-none-any.whl", hash = "sha256:6936baf0ccdc5d9092abd5c2ab628cedcdae9bb72edfdf730f7aaac19118855a", size = 176544, upload-time = "2026-09-28T18:21:21.957Z" }, +] + [[package]] name = "cycler" version = "0.12.1" @@ -1745,28 +2181,85 @@ wheels = [ name = "datasets" version = "5.0.1" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform != 'darwin'", +] dependencies = [ - { name = "dill" }, - { name = "filelock", version = "4.1.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "filelock", version = "4.2.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "fsspec", extra = ["http"], marker = "extra == 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "httpx" }, - { name = "huggingface-hub" }, - { name = "multiprocess" }, - { name = "numpy" }, - { name = "packaging" }, - { name = "pandas" }, - { name = "pyarrow" }, - { name = "pyyaml" }, - { name = "requests" }, - { name = "tqdm" }, - { name = "xxhash" }, + { name = "dill", marker = "python_full_version < '3.11'" }, + { name = "filelock", version = "4.1.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "fsspec", extra = ["http"], marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "httpx", marker = "python_full_version < '3.11'" }, + { name = "huggingface-hub", version = "1.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "multiprocess", marker = "python_full_version < '3.11'" }, + { name = "numpy", marker = "python_full_version < '3.11'" }, + { name = "packaging", marker = "python_full_version < '3.11'" }, + { name = "pandas", marker = "python_full_version < '3.11'" }, + { name = "pyarrow", version = "23.0.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "pyyaml", marker = "python_full_version < '3.11'" }, + { name = "requests", marker = "python_full_version < '3.11'" }, + { name = "tqdm", marker = "python_full_version < '3.11'" }, + { name = "xxhash", marker = "python_full_version < '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/0a/5b/836516269d4f618efe621661cfb6f9acc57e6f95265db3efaee48a5ffe04/datasets-5.0.1.tar.gz", hash = "sha256:ce22bb851efd7494f08aad33b940803784434f6e77763d00679a0dc45fcf686a", size = 641498, upload-time = "2026-07-28T11:09:12.016Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/44/0b/98fc6eb83333508ca5f44c52b3e287ea8137a0ad582714e2cbc67a02154b/datasets-5.0.1-py3-none-any.whl", hash = "sha256:9fbf73688f8c18f7529b4fe592abd04015f81d1e58001e4bac73ffb2b39d7cc4", size = 559079, upload-time = "2026-07-28T11:09:10.266Z" }, ] +[[package]] +name = "datasets" +version = "5.1.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "dill", marker = "python_full_version >= '3.11'" }, + { name = "filelock", version = "4.2.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "fsspec", extra = ["http"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "huggingface-hub", version = "1.33.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "huggingface-hub", version = "2.3.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "multiprocess", marker = "python_full_version >= '3.11'" }, + { name = "numpy", marker = "python_full_version >= '3.11'" }, + { name = "packaging", marker = "python_full_version >= '3.11'" }, + { name = "pandas", marker = "python_full_version >= '3.11'" }, + { name = "pyarrow", version = "26.0.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "pyyaml", marker = "python_full_version >= '3.11'" }, + { name = "tqdm", marker = "python_full_version >= '3.11'" }, + { name = "xxhash", marker = "python_full_version >= '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/17/e6/b5d8c77f13cc6269660b0f2691e8875777b9d94972e281b7b2c25f25bd9d/datasets-5.1.0.tar.gz", hash = "sha256:1561347daa8841b51351ef2ad1647b4ee4af406852c0f7bd1df587ac1ca7888e", size = 687306, upload-time = "2026-10-05T15:32:32.053Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/24/cc/7f624629846c709ef1c68aa61bfb8d0ae2638cbe265104a1c2f7a630949e/datasets-5.1.0-py3-none-any.whl", hash = "sha256:86fd08b280485aab82894d2c81842c4151d93b575efea0054315104cb7bc3cec", size = 604432, upload-time = "2026-10-05T15:32:29.892Z" }, +] + [[package]] name = "decorator" version = "5.2.1" @@ -1833,8 +2326,8 @@ name = "depyf" version = "0.20.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "astor" }, - { name = "dill" }, + { name = "astor", marker = "python_full_version >= '3.11'" }, + { name = "dill", marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/88/35/83fb0178212279aa0af031031905804c6de5618435d229f41ed21bb9ad2c/depyf-0.20.0.tar.gz", hash = "sha256:fb7683bd72c44f67b56029df2c47721e9a02ffa4d7b19095f1c54c4ebf797a98", size = 6168761, upload-time = "2025-10-13T12:33:38.589Z" } wheels = [ @@ -1918,8 +2411,8 @@ name = "email-validator" version = "2.3.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "dnspython" }, - { name = "idna" }, + { name = "dnspython", marker = "python_full_version >= '3.11'" }, + { name = "idna", version = "3.20", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/f5/22/900cb125c76b7aaa450ce02fd727f452243f2e91a61af068b40adba60ea9/email_validator-2.3.0.tar.gz", hash = "sha256:9fc05c37f2f6cf439ff414f8fc46d917929974a82244c20eb10231ba60c54426", size = 51238, upload-time = "2025-08-26T13:09:06.831Z" } wheels = [ @@ -1952,13 +2445,13 @@ name = "fastapi" version = "0.135.3" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "annotated-doc" }, - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "annotated-doc", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-sglang') or (python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "starlette" }, + { name = "starlette", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-inspection" }, + { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-inspection", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/f7/e6/7adb4c5fa231e82c35b8f5741a9f2d055f520c29af5546fd70d3e8e1cd2e/fastapi-0.135.3.tar.gz", hash = "sha256:bd6d7caf1a2bdd8d676843cdcd2287729572a1ef524fc4d65c17ae002a1be654", size = 396524, upload-time = "2026-04-01T16:23:58.188Z" } wheels = [ @@ -1967,14 +2460,14 @@ wheels = [ [package.optional-dependencies] standard = [ - { name = "email-validator" }, - { name = "fastapi-cli", extra = ["standard"], marker = "extra == 'extra-7-vec-inf-vllm'" }, - { name = "httpx" }, - { name = "jinja2" }, - { name = "pydantic-extra-types" }, - { name = "pydantic-settings" }, - { name = "python-multipart" }, - { name = "uvicorn", extra = ["standard"], marker = "extra == 'extra-7-vec-inf-vllm'" }, + { name = "email-validator", marker = "python_full_version >= '3.11'" }, + { name = "fastapi-cli", extra = ["standard"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "httpx", marker = "python_full_version >= '3.11'" }, + { name = "jinja2", marker = "python_full_version >= '3.11'" }, + { name = "pydantic-extra-types", marker = "python_full_version >= '3.11'" }, + { name = "pydantic-settings", marker = "python_full_version >= '3.11'" }, + { name = "python-multipart", marker = "python_full_version >= '3.11'" }, + { name = "uvicorn", extra = ["standard"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] [[package]] @@ -1982,10 +2475,9 @@ name = "fastapi-cli" version = "0.0.24" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "rich-toolkit" }, - { name = "tomli", marker = "python_full_version < '3.11'" }, - { name = "typer" }, - { name = "uvicorn", extra = ["standard"], marker = "extra == 'extra-7-vec-inf-vllm'" }, + { name = "rich-toolkit", marker = "python_full_version >= '3.11'" }, + { name = "typer", marker = "python_full_version >= '3.11'" }, + { name = "uvicorn", extra = ["standard"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/6e/58/74797ae9e4610cfa0c6b34c8309096d3b20bb29be3b8b5fbf1004d10fa5f/fastapi_cli-0.0.24.tar.gz", hash = "sha256:1afc9c9e21d7ebc8a3ca5e31790cd8d837742be7e4f8b9236e99cb3451f0de00", size = 19043, upload-time = "2026-02-24T10:45:10.476Z" } wheels = [ @@ -1994,8 +2486,8 @@ wheels = [ [package.optional-dependencies] standard = [ - { name = "fastapi-cloud-cli" }, - { name = "uvicorn", extra = ["standard"], marker = "extra == 'extra-7-vec-inf-vllm'" }, + { name = "fastapi-cloud-cli", marker = "python_full_version >= '3.11'" }, + { name = "uvicorn", extra = ["standard"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] [[package]] @@ -2003,15 +2495,15 @@ name = "fastapi-cloud-cli" version = "0.16.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "fastar" }, - { name = "httpx" }, - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, extra = ["email"], marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "fastar", marker = "python_full_version >= '3.11'" }, + { name = "httpx", marker = "python_full_version >= '3.11'" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, extra = ["email"], marker = "(python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, extra = ["email"], marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "rich-toolkit" }, - { name = "rignore" }, - { name = "sentry-sdk" }, - { name = "typer" }, - { name = "uvicorn", extra = ["standard"], marker = "extra == 'extra-7-vec-inf-vllm'" }, + { name = "rich-toolkit", marker = "python_full_version >= '3.11'" }, + { name = "rignore", marker = "python_full_version >= '3.11'" }, + { name = "sentry-sdk", marker = "python_full_version >= '3.11'" }, + { name = "typer", marker = "python_full_version >= '3.11'" }, + { name = "uvicorn", extra = ["standard"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/22/70/ca14fae57a221610d3e2e3dfad2b6e97ee31fcafaa36f90a2158d57e9a73/fastapi_cloud_cli-0.16.1.tar.gz", hash = "sha256:33b552c4ad46cd33823ef53f93b8b7813db2306c80c1cbcfa4d72067c99b26ab", size = 46193, upload-time = "2026-04-08T09:12:54.151Z" } wheels = [ @@ -2136,18 +2628,37 @@ wheels = [ [[package]] name = "fastsafetensors" -version = "0.3.1" -source = { registry = "https://pypi.org/simple" } -dependencies = [ - { name = "typer" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/d2/69/e34a1e86a02b255896c57263bf0dfbae45b4708fd609b937f783c2202e7b/fastsafetensors-0.3.1.tar.gz", hash = "sha256:b7eb039a564d77280d17e5d63b27e9963ba5158ad02d2a3c1772c62072a81a53", size = 55665, upload-time = "2026-05-06T08:48:59.125Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/0a/77/0913cd907585085b2f10f2d7eef2aa52c9445241dc66238de25ee24c8241/fastsafetensors-0.3.1-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:acc0971b006f381598d32c5f7aae3a07d9a10d87c701f26b65215083e15a4733", size = 1808292, upload-time = "2026-05-06T08:48:51.12Z" }, - { url = "https://files.pythonhosted.org/packages/e8/67/eaa10409a526242253926fe6981c652dfdb8aa4ec0d4cba4077a9376a1fd/fastsafetensors-0.3.1-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:32cf4b531b5d77de41d777106ea69036a16bdea80062646dabc93a11b4cf88ee", size = 1828304, upload-time = "2026-05-06T08:48:53.15Z" }, - { url = "https://files.pythonhosted.org/packages/6f/50/909871d673bacd6dfc7fee5e59bcd4ec9fbd19775bafe567ad236a3adced/fastsafetensors-0.3.1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:ac76f33e47959b7c31658fbbda1805df7540819828a3ce6a94eb34b4db0b1fa7", size = 1854825, upload-time = "2026-05-06T08:48:54.452Z" }, - { url = "https://files.pythonhosted.org/packages/84/f0/a9ec204e866b52ce323bafc31f4dbf15581b3c998d597ec53c62012716ff/fastsafetensors-0.3.1-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5c9f8d3e18969090b5d66581cb6e9f6c9057158c151ed87e9c309dead5c64442", size = 1854843, upload-time = "2026-05-06T08:48:55.921Z" }, - { url = "https://files.pythonhosted.org/packages/73/aa/00acc9e8f8209f513bf504371f0c40f4073b90240ef3e73507461d2390b9/fastsafetensors-0.3.1-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:348814ad027891317ad8aff90a3c62f8f704a0861d089ea33cff088980602355", size = 1852255, upload-time = "2026-05-06T08:48:57.442Z" }, +version = "0.5.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "typer", marker = "python_full_version >= '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/9a/92/1deed9e83a70391666e906c71a6561f23e129b79fb5de05450d77488c258/fastsafetensors-0.5.0.tar.gz", hash = "sha256:2532cd58c779035eeadea4f8da8cd4340ba1d4a1a04753946e4f89aed2043c13", size = 128319, upload-time = "2026-10-17T14:49:08.936Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/ba/45/dd51c31243866815d5893da80fe8662f10eebf7823807c7c5263d5d668c5/fastsafetensors-0.5.0-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6f4063a7b351c365adb0a897a7cf034d83ffbefec50979fa637f40ae10b5a519", size = 2638444, upload-time = "2026-10-17T14:48:29.272Z" }, + { url = "https://files.pythonhosted.org/packages/45/b8/4f5b40aa9e7df6010fa3ba1fd0d36fcb2d0d3e74b96caa1d687710d34a7b/fastsafetensors-0.5.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:c36fd85d0f28dd2ab335eeb3846d1751935babfae336c4b5a0dccec3e93b93cd", size = 2670743, upload-time = "2026-10-17T14:48:31.586Z" }, + { url = "https://files.pythonhosted.org/packages/cf/8a/2cb9d8ac02372622acfb93e740a3d43ccc25a5d3c1d887667ce955a5a1d7/fastsafetensors-0.5.0-cp310-cp310-win_amd64.whl", hash = "sha256:4096189c8d5f135f671c0e68906eecfcb3e154d5e255171f514de46e76d596d7", size = 508001, upload-time = "2026-10-17T14:48:32.896Z" }, + { url = "https://files.pythonhosted.org/packages/4f/5f/c08e4d86e52827c2f6cdf4ad9c4299a660ce414c1fe297ccdb6c42ab4c11/fastsafetensors-0.5.0-cp310-cp310-win_arm64.whl", hash = "sha256:59c57798ef435a866538e481682efb166bd5a1c3b9e73274ac43a07713beda5c", size = 710897, upload-time = "2026-10-17T14:48:34.279Z" }, + { url = "https://files.pythonhosted.org/packages/89/7b/fb2c2a369da1f37c502b3364bae95592139efc418762be6d5e71c1ac1605/fastsafetensors-0.5.0-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:c2bf413b6569199db5fa436c0037dfa7e887e17b89ab49faf393398579646146", size = 2657157, upload-time = "2026-10-17T14:48:36.635Z" }, + { url = "https://files.pythonhosted.org/packages/2c/0e/24d7d618c20cbcc8167c502b5194a8c98ad62c75f32267ff42c12c59b6d0/fastsafetensors-0.5.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:33081163451a02e1ab80b1ba35459632e584bac79f301413a99fccb3699df426", size = 2688920, upload-time = "2026-10-17T14:48:38.353Z" }, + { url = "https://files.pythonhosted.org/packages/16/7d/2df4f1354b7eb42d3649da586f625b7ab29cdeef1fdba4e318e356efa45c/fastsafetensors-0.5.0-cp311-cp311-win_amd64.whl", hash = "sha256:2e2de360d7ab9d9c5bebe8d8a53b32dc1311ee70201f4b137515779125c89a4c", size = 509669, upload-time = "2026-10-17T14:48:39.718Z" }, + { url = "https://files.pythonhosted.org/packages/ff/b4/eae9a86554327991dea7b204ccc1efaefa8c4fecb83af9d92057add3295b/fastsafetensors-0.5.0-cp311-cp311-win_arm64.whl", hash = "sha256:1363303033fbb504146411baa60e4cf42803fe56a6f807cf85e98e0a238d9910", size = 711642, upload-time = "2026-10-17T14:48:41.529Z" }, + { url = "https://files.pythonhosted.org/packages/df/6d/01c61ec9f10decacf611d24287aaa12dffc8bb9efca006c1cf0f129beece/fastsafetensors-0.5.0-cp312-cp312-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:dff56f2e405948404703c5e14ef349b9b613def88ffbe6542602c592cad6dd94", size = 2689215, upload-time = "2026-10-17T14:48:43.368Z" }, + { url = "https://files.pythonhosted.org/packages/3a/e4/b3b9feac988f3b401789ab6e4f810487e2a844a5fc8dd3021e7a21f930ff/fastsafetensors-0.5.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:cf8085203d83abbf2ea37edfb4b007cf1de521ab1cd29b46935f611d68c80d7d", size = 2731897, upload-time = "2026-10-17T14:48:45.131Z" }, + { url = "https://files.pythonhosted.org/packages/c3/df/6c03ecf4a78d3771554c09d407d0f7669b4ec7320689af6d0a4552633f36/fastsafetensors-0.5.0-cp312-cp312-win_amd64.whl", hash = "sha256:4efadb3d1eb973845a64facee2dbd186d26ae551f8d0ca744fcec15934c90980", size = 511148, upload-time = "2026-10-17T14:48:46.591Z" }, + { url = "https://files.pythonhosted.org/packages/a5/d3/5a33ba57b1c0780ddbfb184a6f2ca8e7308159b2fe236d0c70510931d276/fastsafetensors-0.5.0-cp312-cp312-win_arm64.whl", hash = "sha256:ad0635c401c02da0945d3619172d9faddc1b54632def19fe039118c7b0cf4a5b", size = 708141, upload-time = "2026-10-17T14:48:48.019Z" }, + { url = "https://files.pythonhosted.org/packages/25/f1/80365dab6dfadd159e3987b38277216ea893ef0a956f78db77295d38ab7a/fastsafetensors-0.5.0-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8d5031896eb76bda48d3cc2d07318f3053bbb7640c6f70a9a6a783deba4ebe7d", size = 2689067, upload-time = "2026-10-17T14:48:49.678Z" }, + { url = "https://files.pythonhosted.org/packages/cf/71/e7f09de184c45b884040b14b285f42cacd9191bb624fd77e666a449b56d1/fastsafetensors-0.5.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:42704d63db61498afcd04cc18c4e35bfde461987be3efd28c54c02042eb54003", size = 2731145, upload-time = "2026-10-17T14:48:51.433Z" }, + { url = "https://files.pythonhosted.org/packages/e2/dd/d26e580eb42d771a135268e2afb0f737f67a0c1467b79d75dd611b10a2bc/fastsafetensors-0.5.0-cp313-cp313-win_amd64.whl", hash = "sha256:302e074c0c7a0f0287297ce19789561dd7b670880619d3c4ac35887fd6157f38", size = 511169, upload-time = "2026-10-17T14:48:53.278Z" }, + { url = "https://files.pythonhosted.org/packages/8a/04/df7b87b14051f1b85b492b1dbb569a9700d78cc6f6023b1ac4236ebea981/fastsafetensors-0.5.0-cp313-cp313-win_arm64.whl", hash = "sha256:43f729472a5fc9d8ce4d1db25c724b953dec76edf948dcdc2ad4c303abc4cd80", size = 708140, upload-time = "2026-10-17T14:48:54.928Z" }, + { url = "https://files.pythonhosted.org/packages/b2/30/d30cd3d75d8fd9908aab21c1d3f2ba6cdf89c182ce9e5b50bc6df43c28e8/fastsafetensors-0.5.0-cp314-cp314-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:cdf6d9ac44027372e344fcc3c2b8ad4c9ecc4b59269f5db1589f5fd23e6de218", size = 2687536, upload-time = "2026-10-17T14:48:56.296Z" }, + { url = "https://files.pythonhosted.org/packages/3e/6c/976bab2021c78a4e4a4748b9db4bd3ec356a53123a216e0748a1b9212e95/fastsafetensors-0.5.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:cff880ecd135818bd34296290ae4b98e2774a4c26d56bb772fc9afebc2fc854e", size = 2727287, upload-time = "2026-10-17T14:48:57.859Z" }, + { url = "https://files.pythonhosted.org/packages/ff/e3/3e53bb109f23f5be5b2355fca8f5cdaf63d883053d102b48a60c6312ca08/fastsafetensors-0.5.0-cp314-cp314-win_amd64.whl", hash = "sha256:c950d4825c87f21e91240ce7b1bd5b9f89c04e037396abba4810c4b506271676", size = 522725, upload-time = "2026-10-17T14:48:59.283Z" }, + { url = "https://files.pythonhosted.org/packages/4a/68/050bdfb9b710c47caf9fcc6b7269782c5a7fe3d0fd9578368634eb505392/fastsafetensors-0.5.0-cp314-cp314-win_arm64.whl", hash = "sha256:34e90f37ee587260559211eb6465be92891663ac38064692e02b06b07c4d20a3", size = 727345, upload-time = "2026-10-17T14:49:00.896Z" }, + { url = "https://files.pythonhosted.org/packages/84/fa/3a8d9bbc3b2b4b126bffb69f9f2cd2a870ff46c1b381c7461fa813a437d5/fastsafetensors-0.5.0-cp315-cp315-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:c0eec60e95ebe08a05db2d3f7ab320eeec559c0559122f39bbc7eb30dcea3f35", size = 2688233, upload-time = "2026-10-17T14:49:02.566Z" }, + { url = "https://files.pythonhosted.org/packages/42/74/51411b9369c752a45b6cb2b3e431bdbd5accc11da8417485d046defeff11/fastsafetensors-0.5.0-cp315-cp315-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f4bf63b147c430d23eb86693e80bf6484f3870d4b4c5671c3f1a34f45c37b3fa", size = 2733245, upload-time = "2026-10-17T14:49:04.331Z" }, + { url = "https://files.pythonhosted.org/packages/46/81/0e97da1f585062164c815ac9b57f7aedf85e6c27dcb39dd47cea19e32dd3/fastsafetensors-0.5.0-cp315-cp315-win_amd64.whl", hash = "sha256:6111b5dbe857037eaaff7dc8973d3f61d6aec217759bdba2750fb2e26b1aeae0", size = 522725, upload-time = "2026-10-17T14:49:06.128Z" }, + { url = "https://files.pythonhosted.org/packages/de/94/224cb8b8e781e28b56da0715450fdd0381ad109bce6f540131e5c3931faa/fastsafetensors-0.5.0-cp315-cp315-win_arm64.whl", hash = "sha256:75fbf9f8559e1b0a8ab823741e56105ab784ec092ef25d118c18f726dd5b888c", size = 727313, upload-time = "2026-10-17T14:49:07.552Z" }, ] [[package]] @@ -2171,23 +2682,33 @@ name = "filelock" version = "4.2.4" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -2205,8 +2726,8 @@ dependencies = [ { name = "apache-tvm-ffi", version = "0.1.9", source = { registry = "https://pypi.org/simple" } }, { name = "einops" }, { name = "nvidia-cutlass-dsl", version = "4.5.1", source = { registry = "https://pypi.org/simple" } }, - { name = "quack-kernels" }, - { name = "torch" }, + { name = "quack-kernels", version = "0.4.1", source = { registry = "https://pypi.org/simple" } }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" } }, { name = "torch-c-dlpack-ext" }, { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, @@ -2220,44 +2741,10 @@ wheels = [ name = "flashinfer-cubin" version = "0.6.11.post1" source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'darwin'", - "python_full_version >= '3.14' and sys_platform != 'darwin'", - "python_full_version == '3.13.*' and sys_platform == 'darwin'", - "python_full_version == '3.13.*' and sys_platform != 'darwin'", - "python_full_version == '3.12.*' and sys_platform == 'darwin'", - "python_full_version == '3.12.*' and sys_platform != 'darwin'", - "python_full_version == '3.11.*' and sys_platform == 'darwin'", - "python_full_version == '3.11.*' and sys_platform != 'darwin'", - "python_full_version < '3.11' and sys_platform == 'darwin'", - "python_full_version < '3.11' and sys_platform != 'darwin'", -] wheels = [ { url = "https://files.pythonhosted.org/packages/56/0c/80dad211d424c3f25199ccd9bb1913c3e2d7378b5cd3dbcd2f75a635b6dd/flashinfer_cubin-0.6.11.post1-py3-none-any.whl", hash = "sha256:1eb801fc80b5576760d356e31eb452d05ab240f44185f539331bd5f2236e504a", size = 360908523, upload-time = "2026-05-13T01:31:43.386Z" }, ] -[[package]] -name = "flashinfer-cubin" -version = "0.6.11.post2" -source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32'", - "python_full_version >= '3.14' and sys_platform == 'emscripten'", - "python_full_version >= '3.14' and sys_platform == 'darwin'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", - "python_full_version == '3.13.*' and sys_platform == 'darwin'", - "python_full_version == '3.13.*' and sys_platform != 'darwin'", - "python_full_version == '3.12.*' and sys_platform == 'darwin'", - "python_full_version == '3.12.*' and sys_platform != 'darwin'", - "python_full_version == '3.11.*' and sys_platform == 'darwin'", - "python_full_version == '3.11.*' and sys_platform != 'darwin'", - "python_full_version < '3.11' and sys_platform == 'darwin'", - "python_full_version < '3.11' and sys_platform != 'darwin'", -] -wheels = [ - { url = "https://files.pythonhosted.org/packages/29/96/da75a9f61c64c87b16baa339fc8216a6c3743c5d263c555fded30fcbe6f7/flashinfer_cubin-0.6.11.post2-py3-none-any.whl", hash = "sha256:eb01c2801ee31d145bbf7afb2c223150333e602c8208216017b0190b1087b990", size = 360908523, upload-time = "2026-05-14T04:57:41.355Z" }, -] - [[package]] name = "flashinfer-python" version = "0.6.11.post1" @@ -2276,18 +2763,20 @@ resolution-markers = [ ] dependencies = [ { name = "apache-tvm-ffi", version = "0.1.9", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, - { name = "click", marker = "extra == 'extra-7-vec-inf-sglang'" }, - { name = "cuda-tile", marker = "extra == 'extra-7-vec-inf-sglang'" }, + { name = "click", version = "8.3.2", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cuda-tile", version = "1.3.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cuda-tile", version = "1.7.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "einops", marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "ninja", marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "numpy", marker = "extra == 'extra-7-vec-inf-sglang'" }, - { name = "nvidia-cudnn-frontend", marker = "extra == 'extra-7-vec-inf-sglang'" }, + { name = "nvidia-cudnn-frontend", version = "1.18.0", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "nvidia-cutlass-dsl", version = "4.5.1", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "nvidia-ml-py", marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "packaging", marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "requests", marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "tabulate", marker = "extra == 'extra-7-vec-inf-sglang'" }, - { name = "torch", marker = "extra == 'extra-7-vec-inf-sglang'" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "tqdm", marker = "extra == 'extra-7-vec-inf-sglang'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/f1/4b/d32491a43d21de0a7a780fb6ce00749d259cd685ddbc1bee7932d8a3a513/flashinfer_python-0.6.11.post1.tar.gz", hash = "sha256:b16ac9a4f229a6a0d875a2fd76dfd935d5c62e4a99ea7b128fe4b457530a5aed", size = 9215625, upload-time = "2026-05-13T01:31:31.613Z" } @@ -2300,44 +2789,83 @@ name = "flashinfer-python" version = "0.6.11.post2" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", ] dependencies = [ - { name = "apache-tvm-ffi", version = "0.1.9", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-vllm'" }, - { name = "apache-tvm-ffi", version = "0.1.10", source = { registry = "https://pypi.org/simple" }, marker = "(extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "click", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "cuda-tile", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "einops", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "ninja", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "numpy", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "nvidia-cudnn-frontend", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "nvidia-cutlass-dsl", version = "4.5.2", source = { registry = "https://pypi.org/simple" }, marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "nvidia-ml-py", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "packaging", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "requests", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "tabulate", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "torch", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "tqdm", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, + { name = "apache-tvm-ffi", version = "0.1.10", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "click", version = "8.3.2", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cuda-tile", version = "1.3.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "einops", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "ninja", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "numpy", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cudnn-frontend", version = "1.18.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cutlass-dsl", version = "4.5.2", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-ml-py", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "packaging", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "requests", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tabulate", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tqdm", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/db/53/dbf2157f2bbb96d6f7a6891cf6abfb2e6e18963760a0c53e96c2de5c59db/flashinfer_python-0.6.11.post2.tar.gz", hash = "sha256:e9fdac56aea9f0f58a4e69b0645c54993760d3cc6c7bf5c2df4ce5a0aecc7953", size = 9248515, upload-time = "2026-05-14T04:57:32.83Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/c7/bc/518b092473f37d904ae07766ad37c772b93da13ea788777b22a80c3f1a7c/flashinfer_python-0.6.11.post2-py3-none-any.whl", hash = "sha256:550cbdb760f9f7ec0e42055e06636b9489d05f1a38989cafd77e6eb820de0138", size = 13746417, upload-time = "2026-05-14T04:57:30.25Z" }, ] +[[package]] +name = "flashinfer-python" +version = "0.7.0.post1" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "apache-tvm-ffi", version = "0.1.11", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cuda-python", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cuda-tile", version = "1.7.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "einops", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nccl-extensions", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nccl4py", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "ninja", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "numpy", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cudnn-frontend", version = "1.31.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cutlass-dsl", version = "4.8.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-ml-py", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "packaging", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "requests", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tabulate", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tqdm", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/14/b0/2f27f2645056b0e671b3bcf8a934e73ed89ca5c5df9ebe651f21c0d218e2/flashinfer_python-0.7.0.post1.tar.gz", hash = "sha256:2e67b38a9af1d4eedff96f7f5d79c6a4d9f0f97cb4a30f71e4ffd717f1ad9af4", size = 26061698, upload-time = "2026-09-29T23:34:14.797Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/00/7d/16237087a2af99086ef4ae9808dff9b24f0a4db4cfd673233836821a32e0/flashinfer_python-0.7.0.post1-py3-none-any.whl", hash = "sha256:c7adf826568d61fc1b7d3aadd4cae387a35a138bfc08e2deca2f34ecfa280716", size = 33659854, upload-time = "2026-09-29T23:34:11.137Z" }, +] + [[package]] name = "fonttools" version = "4.62.1" @@ -2527,7 +3055,7 @@ wheels = [ [package.optional-dependencies] http = [ - { name = "aiohttp" }, + { name = "aiohttp", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, ] [[package]] @@ -2562,7 +3090,7 @@ name = "gitdb" version = "4.0.12" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "smmap" }, + { name = "smmap", marker = "python_full_version < '3.11' or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/72/94/63b0fc47eb32792c7ba1fe1b694daec9a63620db1e313033d18140c2320a/gitdb-4.0.12.tar.gz", hash = "sha256:5ef71f855d191a3326fcfbc0d5da835f26b13fbcba60c32c21091c349ffdb571", size = 394684, upload-time = "2025-01-02T07:20:46.413Z" } wheels = [ @@ -2574,7 +3102,7 @@ name = "gitpython" version = "3.2.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "gitdb" }, + { name = "gitdb", marker = "python_full_version < '3.11' or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/c0/ff/fee293f1ec3e1b64db6eb270e00964f29c086d769201474cedb8d1608930/gitpython-3.2.1.tar.gz", hash = "sha256:0f1229ec5c057bdabdeeda8bda4f17bda679ab33dbccc00f24e6a480553c8613", size = 241688, upload-time = "2026-10-14T12:20:20.589Z" } wheels = [ @@ -2589,7 +3117,8 @@ dependencies = [ { name = "google-auth" }, { name = "googleapis-common-protos" }, { name = "proto-plus" }, - { name = "protobuf" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "requests" }, ] sdist = { url = "https://files.pythonhosted.org/packages/16/ce/502a57fb0ec752026d24df1280b162294b22a0afb98a326084f9a979138b/google_api_core-2.30.3.tar.gz", hash = "sha256:e601a37f148585319b26db36e219df68c5d07b6382cff2d580e83404e44d641b", size = 177001, upload-time = "2026-04-10T00:41:28.035Z" } @@ -2615,7 +3144,8 @@ name = "googleapis-common-protos" version = "1.74.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "protobuf" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/20/18/a746c8344152d368a5aac738d4c857012f2c5d1fd2eac7e17b647a7861bd/googleapis_common_protos-1.74.0.tar.gz", hash = "sha256:57971e4eeeba6aad1163c1f0fc88543f965bb49129b8bb55b2b7b26ecab084f1", size = 151254, upload-time = "2026-04-02T21:23:26.679Z" } wheels = [ @@ -2732,28 +3262,82 @@ wheels = [ name = "grpcio-health-checking" version = "1.81.1" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform != 'darwin'", +] dependencies = [ - { name = "grpcio" }, - { name = "protobuf" }, + { name = "grpcio", marker = "python_full_version < '3.11'" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/65/46/6b6678b5a922765ae7637205bb6d0618a4da8b35f2ce6116f8bcff262370/grpcio_health_checking-1.81.1.tar.gz", hash = "sha256:ecc61480e25058a4a04e11e4ab6900ad7439b32e60a8ce4ece7d9f219221c85d", size = 17107, upload-time = "2026-06-11T12:58:49.632Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/6e/87/174bbfb5613794862c45b5cdf567fcfb478ad5a3a7b594e11d998656b2f9/grpcio_health_checking-1.81.1-py3-none-any.whl", hash = "sha256:cbc6a4171825ec64389de2f062d296ba129a5c27eefd0dd55fa837909184bdf9", size = 19120, upload-time = "2026-06-11T12:58:38.008Z" }, ] +[[package]] +name = "grpcio-health-checking" +version = "1.83.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and sys_platform != 'darwin'", +] +dependencies = [ + { name = "grpcio", marker = "python_full_version >= '3.11'" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/cd/84/5db47abe1b8b946e8edcfcf7fb2116c7a4ab839e23126680a984c88eb424/grpcio_health_checking-1.83.0.tar.gz", hash = "sha256:ad6bc4d5a1103ad704d25ccd82def300dfa27996b185cab3e4cceeef2c6867d4", size = 17061, upload-time = "2026-07-23T15:24:23.019Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/c6/9c/8922f2a529bf49361b4f70d6d07adfadc07d1a93eb5ee07f6f66a54445e8/grpcio_health_checking-1.83.0-py3-none-any.whl", hash = "sha256:7d8b47a5bfbc699d4aee0fc7a27f5d0265eb23a6a64db5ac2d8b749a0f8a9911", size = 19085, upload-time = "2026-07-23T15:24:07.792Z" }, +] + [[package]] name = "grpcio-reflection" version = "1.81.1" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform != 'darwin'", +] dependencies = [ - { name = "grpcio" }, - { name = "protobuf" }, + { name = "grpcio", marker = "python_full_version < '3.11'" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/c0/53/7bd579cb35ea5895cb1ef38066e2336324432b3d3eaa21f52f6b94457c2d/grpcio_reflection-1.81.1.tar.gz", hash = "sha256:3d7160000f5fdd0e241f9b1a3d402f15ebcd5f281be105b374a025f38c6434a8", size = 19203, upload-time = "2026-06-11T12:58:47.865Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/3b/10/5152fbc98f6b5b4f54cdf97c585d19fe962c265774692e446ad6b4f4950c/grpcio_reflection-1.81.1-py3-none-any.whl", hash = "sha256:d82fc4ac39dd5dcfd63e577075f6a68eac4a765e66a453914bbf57caf35665d0", size = 22906, upload-time = "2026-06-11T12:58:03.183Z" }, ] +[[package]] +name = "grpcio-reflection" +version = "1.83.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and sys_platform != 'darwin'", +] +dependencies = [ + { name = "grpcio", marker = "python_full_version >= '3.11'" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/e5/6e/e46a65398fe4289b62f46266234321fcfb9ff96fcd69c872440c2bfaf607/grpcio_reflection-1.83.0.tar.gz", hash = "sha256:6a2a30a462ac2c3c6d149734a8fe655fa7617c17fa2cda9949006f3bf07f5b3e", size = 19203, upload-time = "2026-07-23T15:24:26.163Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/8c/f2/21a1bbe4d35efbfb301dca50406bf643a503c041c8bafc53686c7ed180a2/grpcio_reflection-1.83.0-py3-none-any.whl", hash = "sha256:9d4151f68f18a85aa5b0eff5d493ad13f9421f96c8d9c01c840b353dc979be93", size = 22910, upload-time = "2026-07-23T15:23:37.94Z" }, +] + [[package]] name = "h11" version = "0.16.0" @@ -2767,6 +3351,13 @@ wheels = [ name = "hf-xet" version = "1.4.3" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] sdist = { url = "https://files.pythonhosted.org/packages/53/92/ec9ad04d0b5728dca387a45af7bc98fbb0d73b2118759f5f6038b61a57e8/hf_xet-1.4.3.tar.gz", hash = "sha256:8ddedb73c8c08928c793df2f3401ec26f95be7f7e516a7bee2fbb546f6676113", size = 670477, upload-time = "2026-03-31T22:40:07.874Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/72/43/724d307b34e353da0abd476e02f72f735cdd2bc86082dee1b32ea0bfee1d/hf_xet-1.4.3-cp313-cp313t-macosx_10_12_x86_64.whl", hash = "sha256:7551659ba4f1e1074e9623996f28c3873682530aee0a846b7f2f066239228144", size = 3800935, upload-time = "2026-03-31T22:39:49.618Z" }, @@ -2795,6 +3386,52 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/8a/7c/44314ecd0e89f8b2b51c9d9e5e7a60a9c1c82024ac471d415860557d3cd8/hf_xet-1.4.3-cp37-abi3-win_arm64.whl", hash = "sha256:7c2c7e20bcfcc946dc67187c203463f5e932e395845d098cc2a93f5b67ca0b47", size = 3533664, upload-time = "2026-03-31T22:40:12.152Z" }, ] +[[package]] +name = "hf-xet" +version = "1.7.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +sdist = { url = "https://files.pythonhosted.org/packages/9e/27/06d899ea7bd721d272f84aac98bdb238de98af4cc767a69056d967d68c71/hf_xet-1.7.0.tar.gz", hash = "sha256:d406ec79053c0871817f700c2ac8c36ba0d87f9c34b7458b0f0063bb218b0466", size = 985689, upload-time = "2026-10-06T20:18:43.89Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/9f/7c/3e45174942e6793adde6cba4daa7fb037275cf02a944d9eadfcf9ff33b86/hf_xet-1.7.0-cp314-cp314t-macosx_10_12_x86_64.whl", hash = "sha256:fa029678be1ba7f953c409b0b27bf15cc69cd1c9b3a674fbd78856ebefca1052", size = 3803919, upload-time = "2026-10-06T20:18:09.844Z" }, + { url = "https://files.pythonhosted.org/packages/ff/3a/5e8b363391adcbb002e191dbf924dab31464ea9c45adfeb73502afc36d35/hf_xet-1.7.0-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:57bc157b8b7fe3bee9dcb9af7f3da8de41801c3b31a9ef68a77a33c6a6be382f", size = 3553588, upload-time = "2026-10-06T20:18:13.376Z" }, + { url = "https://files.pythonhosted.org/packages/e5/c2/0d1eaa5da13bbf9c896badc7f380601c7d973a87a6ffb4d100267c4536c1/hf_xet-1.7.0-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:87dab080f8f7d32781c2586904e3603f4e60d09bfc727706c3ae419e0829beeb", size = 4201962, upload-time = "2026-10-06T20:18:16.11Z" }, + { url = "https://files.pythonhosted.org/packages/23/2d/225d5b11a9ca7d31b9470a57f2b2be1a5cef8b84325a2146aeb4589e226c/hf_xet-1.7.0-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:b01fe18dbbd151a2403d2c64ed30dc6547b00d6babab9a617d77c7acdb81ee66", size = 3982978, upload-time = "2026-10-06T20:18:18.092Z" }, + { url = "https://files.pythonhosted.org/packages/93/34/9d681f0e3dac0b5dae0d7dea748429266f24e52415446523f464fbaa828e/hf_xet-1.7.0-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:4ee5e05a627f5ab5bad7a86582277d645556ea1e199903aae19e033a392aa13a", size = 4181558, upload-time = "2026-10-06T20:18:20.082Z" }, + { url = "https://files.pythonhosted.org/packages/de/f0/277f039b7d72027bc2ed277f1b62a2f70f740a5aac2a3e7243e5b6854c5d/hf_xet-1.7.0-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:19c0e64f14175ccb6a1aff69e0d2ab9ec5269a560e6687abaf2b3fa4f73de7cd", size = 4411546, upload-time = "2026-10-06T20:18:21.999Z" }, + { url = "https://files.pythonhosted.org/packages/3d/7f/832d3ddb49326114175b7bcc50daea8565c09fd21ac03a02b211c09fefb7/hf_xet-1.7.0-cp314-cp314t-win_amd64.whl", hash = "sha256:757168feb5679647c0bb13ee5d0faebe799c4dff9051419885a566ebd79f949d", size = 3812809, upload-time = "2026-10-06T20:18:24.288Z" }, + { url = "https://files.pythonhosted.org/packages/3d/c4/310c3c29e5beae7c049e63947bd1923d597883b41c9ec4718589920812c4/hf_xet-1.7.0-cp314-cp314t-win_arm64.whl", hash = "sha256:b91569d5f1b61c34b043687da02c05dd3604f3d329e7868510bf3f7971599006", size = 3646174, upload-time = "2026-10-06T20:18:26.279Z" }, + { url = "https://files.pythonhosted.org/packages/9c/0b/b03be21ffaada749ba0d3197d8aefbf1aa698bac149580421c15239b299e/hf_xet-1.7.0-cp38-abi3-macosx_10_12_x86_64.whl", hash = "sha256:e3e88a7a75d7d95cbee1f37dc31341d6201124cf21c6c4b1dfab8ccba9b09e0f", size = 3796096, upload-time = "2026-10-06T20:18:28.43Z" }, + { url = "https://files.pythonhosted.org/packages/c3/47/a26ebdce7056a61e931f228439bc0ab08cbec239d1690f965e5e637cba79/hf_xet-1.7.0-cp38-abi3-macosx_11_0_arm64.whl", hash = "sha256:59fba37039233c7fcbe196817d6cdcf1b40dfb17b410f229d85b0cf0a1848da4", size = 3560352, upload-time = "2026-10-06T20:18:30.365Z" }, + { url = "https://files.pythonhosted.org/packages/a3/4c/2bf3b66c215d409655f28de1622393dde04c9461280d48c7924bb3b2decd/hf_xet-1.7.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:2814a6e999d13464c4d679b788cc5d784eb5a4edfc638a31f10e9a11ab531ef8", size = 4212180, upload-time = "2026-10-06T20:18:32.292Z" }, + { url = "https://files.pythonhosted.org/packages/49/0c/a2f703a5a78267556e89e03316fa0805c86b72b50829bc67665746e8ebf0/hf_xet-1.7.0-cp38-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:fcfd6c22418e57dd5b3aea649e813b2e2cfb2aebf317b210d90f1fe4b3018b52", size = 3990011, upload-time = "2026-10-06T20:18:34.21Z" }, + { url = "https://files.pythonhosted.org/packages/a4/77/e52e4201b1cbf571530a61cc57f70182045a39a230089ee5f1df182a4de2/hf_xet-1.7.0-cp38-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:80f79dae613ce9e0ea1fd1ae15616ca9ac74aed4c770aabc199c4f03ebecc863", size = 4190628, upload-time = "2026-10-06T20:18:36.062Z" }, + { url = "https://files.pythonhosted.org/packages/6c/dc/03a21b89f118664a0926ff25b0f8e44a519bf22724a6a8fc7a9abbc188b6/hf_xet-1.7.0-cp38-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:0a9e802f33bf50c851abe45fc5380e61f959e2d369647d6742b79ad9d6c27cab", size = 4418814, upload-time = "2026-10-06T20:18:37.888Z" }, + { url = "https://files.pythonhosted.org/packages/4d/59/b35106dfa71b6eef605dc88bd038fe99c7f86fb132a15b60d0bf2f235b2c/hf_xet-1.7.0-cp38-abi3-win_amd64.whl", hash = "sha256:2b7bb5727889b0f2436dbaaad8fc4c3e66b8240d992716989e0c086b4278b1bc", size = 3822644, upload-time = "2026-10-06T20:18:40.052Z" }, + { url = "https://files.pythonhosted.org/packages/48/cd/072313585f74fe9d441e2eb5e0a4703c30586cd709810ea369675f61b74e/hf_xet-1.7.0-cp38-abi3-win_arm64.whl", hash = "sha256:acc3851cf2576a8fb2ae926da863f4efabe21303cf292e9a44332802ab0dcc6a", size = 3662436, upload-time = "2026-10-06T20:18:42.205Z" }, +] + [[package]] name = "httpcore" version = "1.0.9" @@ -2808,6 +3445,19 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/7e/f5/f66802a942d491edb555dd61e3a9961140fd64c90bce1eafd741609d334d/httpcore-1.0.9-py3-none-any.whl", hash = "sha256:2d400746a40668fc9dec9810239072b40b4484b640a8c38fd654a024c7a1bf55", size = 78784, upload-time = "2025-04-24T22:06:20.566Z" }, ] +[[package]] +name = "httpcore2" +version = "2.16.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "h11", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and sys_platform != 'emscripten' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.14' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "truststore", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.14' and sys_platform != 'emscripten' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.14' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/c0/61/afa9d6b7452f9eca73f73d45324eb0bb52f098fd5e6049bf6b8263be9b4e/httpcore2-2.16.0.tar.gz", hash = "sha256:c72cd0413563b17e1cd8354677dff887466f8ec2551679759e93126534428643", size = 69886, upload-time = "2026-10-21T07:09:45.31Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/af/c2/ce417c397afa5181f0e3c6ff01a4d01639414e1adef08730d97571297151/httpcore2-2.16.0-py3-none-any.whl", hash = "sha256:5c4358319f2602d984ad93182c9fee5160f22e9da7babdda870a5e7c4e2dc34f", size = 85070, upload-time = "2026-10-21T07:09:42.802Z" }, +] + [[package]] name = "httptools" version = "0.7.1" @@ -2859,7 +3509,8 @@ dependencies = [ { name = "anyio" }, { name = "certifi" }, { name = "httpcore" }, - { name = "idna" }, + { name = "idna", version = "3.15", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "idna", version = "3.20", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/b1/df/48c586a5fe32a0f01324ee087459e112ebb7224f646c0b5023f5e79e9956/httpx-0.28.1.tar.gz", hash = "sha256:75e98c5f16b0f35b567856f597f06ff2270a374470a5c2392242528e3e3e42fc", size = 141406, upload-time = "2024-12-06T15:37:23.222Z" } wheels = [ @@ -2867,63 +3518,159 @@ wheels = [ ] [[package]] -name = "httpx-sse" -version = "0.4.3" +name = "httpx2" +version = "2.16.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "anyio", marker = "(python_full_version >= '3.11' and sys_platform != 'emscripten' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "httpcore2", marker = "(python_full_version >= '3.11' and sys_platform != 'emscripten' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "httpx2-jsfetch", marker = "(python_full_version >= '3.12' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "idna", version = "3.20", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "truststore", marker = "(python_full_version >= '3.11' and sys_platform != 'emscripten' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.13' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/43/73/5ca2cedf5e4e50423d1c258877865a6b62793633b561c58499f8aeb33b7e/httpx2-2.16.0.tar.gz", hash = "sha256:0d54c98128e8f2ff7fbbde0fca8c359e5f032edeaaaac443cdb7f31f685bcd7f", size = 103116, upload-time = "2026-10-21T07:09:46.348Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/7b/3d/2e4bc4fd1ef9770871669adb348d612a6e4786b303a23f9935b81c48dd60/httpx2-2.16.0-py3-none-any.whl", hash = "sha256:fd2c5d7909e8d86fe92d53020c310ca0e4e431b838c207c5754b9bbf0308d49a", size = 97880, upload-time = "2026-10-21T07:09:44.01Z" }, +] + +[[package]] +name = "httpx2-jsfetch" +version = "1.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/0f/4c/751061ffa58615a32c31b2d82e8482be8dd4a89154f003147acee90f2be9/httpx_sse-0.4.3.tar.gz", hash = "sha256:9b1ed0127459a66014aec3c56bebd93da3c1bc8bb6618c8082039a44889a755d", size = 15943, upload-time = "2025-10-10T21:48:22.271Z" } +sdist = { url = "https://files.pythonhosted.org/packages/cd/c4/0e5636363151a2a1795e0a77617168b9ca438e1748ec05fc9b5687f93d64/httpx2_jsfetch-1.0.tar.gz", hash = "sha256:70a0e3eabfef7cce5ad9c629f7d01ca05e418f586646f4ddf14782e4c1454c60", size = 6872, upload-time = "2026-08-07T00:13:07.492Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/d2/fd/6668e5aec43ab844de6fc74927e155a3b37bf40d7c3790e49fc0406b6578/httpx_sse-0.4.3-py3-none-any.whl", hash = "sha256:0ac1c9fe3c0afad2e0ebb25a934a59f4c7823b60792691f779fad2c5568830fc", size = 8960, upload-time = "2025-10-10T21:48:21.158Z" }, + { url = "https://files.pythonhosted.org/packages/9b/43/832f631d32e4f1211caa2ba368317739fe71f0b8530e4c9d15dc454bac2a/httpx2_jsfetch-1.0-py3-none-any.whl", hash = "sha256:cb916b707601e69a07721aabc8f3f6659be3a6893bc1ff5c6f9e02241df2da32", size = 6382, upload-time = "2026-08-07T00:13:06.567Z" }, ] [[package]] name = "huggingface-hub" version = "1.15.0" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] dependencies = [ - { name = "filelock", version = "4.1.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "filelock", version = "4.2.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "filelock", version = "4.1.0", source = { registry = "https://pypi.org/simple" } }, { name = "fsspec" }, - { name = "hf-xet", marker = "platform_machine == 'AMD64' or platform_machine == 'aarch64' or platform_machine == 'amd64' or platform_machine == 'arm64' or platform_machine == 'x86_64'" }, + { name = "hf-xet", version = "1.4.3", source = { registry = "https://pypi.org/simple" }, marker = "platform_machine == 'AMD64' or platform_machine == 'aarch64' or platform_machine == 'amd64' or platform_machine == 'arm64' or platform_machine == 'x86_64'" }, { name = "httpx" }, { name = "packaging" }, { name = "pyyaml" }, { name = "tqdm" }, { name = "typer" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" } }, ] sdist = { url = "https://files.pythonhosted.org/packages/bb/b6/e22bd20a25299c34b8c5922c1545a6320825b13906eb0f7298edfd034a0b/huggingface_hub-1.15.0.tar.gz", hash = "sha256:28abfdddda3927fd4de6a63cf26ab012498a2c24dae52baf150c5c6edf98a1d5", size = 784100, upload-time = "2026-05-15T11:42:52.149Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/6e/11/0b64cc9024329b76d7547c19a67604a61d21d3ba678a69d1b220c29d5112/huggingface_hub-1.15.0-py3-none-any.whl", hash = "sha256:a4a59af04cbc41a3fe3fec429b171ef994ef8c971eda10136746f408dd4e3744", size = 663602, upload-time = "2026-05-15T11:42:50.487Z" }, ] +[[package]] +name = "huggingface-hub" +version = "1.33.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and sys_platform != 'darwin'", +] +dependencies = [ + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "filelock", version = "4.2.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "fsspec", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "hf-xet", version = "1.7.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 'AMD64' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'amd64' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'arm64' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "httpx", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "packaging", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pyyaml", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tqdm", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/25/2a/484d112c0d8fc5f665d7b65137ac9cdb2953c982391598c3597968a12ee7/huggingface_hub-1.33.0.tar.gz", hash = "sha256:367be21a201db9523eddf8aeac7048f2602c1b308691c97640d5e72ed188007e", size = 1043548, upload-time = "2026-09-24T09:49:29.971Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/fc/16/963096d224b80909432dc16561a615fd33d2d13beef3ce4c63fa25e40867/huggingface_hub-1.33.0-py3-none-any.whl", hash = "sha256:04e434b06e100eddbce9a6e817d72693a7884b10a79bd67ab48080d5c07eb899", size = 846435, upload-time = "2026-09-24T09:49:28.059Z" }, +] + +[[package]] +name = "huggingface-hub" +version = "2.3.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "filelock", version = "4.2.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "fsspec", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "hf-xet", version = "1.7.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 'AMD64' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'ARM64' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'aarch64' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'amd64' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'arm64' and extra != 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and platform_machine == 'x86_64' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "httpx2", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "packaging", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pyyaml", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tqdm", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/46/dd/585639d95f6a987c0bdfb9045f050f4ad67240775644ba7a4fdc5aca3a3b/huggingface_hub-2.3.0.tar.gz", hash = "sha256:0b783b37f200f1d25bef2344a997c7d410f4437fbe9d8fd965f958e75cec7aef", size = 1063410, upload-time = "2026-10-23T08:52:06.594Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/c7/89/5f6ef1162d0323f167b6baab04d1b4bd6a946161a8af6d86f622d9d79aa1/huggingface_hub-2.3.0-py3-none-any.whl", hash = "sha256:5da369600233568a126eb5578754935b2add9446b42d9e2fc0927d56840ac1a5", size = 855161, upload-time = "2026-10-23T08:52:04.78Z" }, +] + [[package]] name = "humming-kernels" -version = "0.1.2" +version = "0.1.17" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "cuda-bindings" }, - { name = "jinja2" }, - { name = "numpy" }, - { name = "nvidia-ml-py" }, - { name = "pyelftools" }, - { name = "safetensors" }, - { name = "tabulate" }, - { name = "torch" }, - { name = "tqdm" }, - { name = "triton" }, + { name = "cuda-bindings", marker = "python_full_version >= '3.11'" }, + { name = "jinja2", marker = "python_full_version >= '3.11'" }, + { name = "numpy", marker = "python_full_version >= '3.11'" }, + { name = "safetensors", version = "0.8.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "tabulate", marker = "python_full_version >= '3.11'" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "tqdm", marker = "python_full_version >= '3.11'" }, + { name = "triton", version = "3.6.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.14' and sys_platform == 'emscripten' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.14' and sys_platform == 'win32' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "triton", version = "3.7.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and sys_platform == 'emscripten' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and python_full_version < '3.14' and sys_platform == 'win32' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/06/f4/e141f45697b7d0d38bfaf8766a7362d8f0136e3cff2620624f24f68e2700/humming_kernels-0.1.2.tar.gz", hash = "sha256:7894c80061c7866591bef12617da720ac4e925636ffc99464af433a5dcb035eb", size = 117251, upload-time = "2026-05-23T16:18:08.084Z" } +sdist = { url = "https://files.pythonhosted.org/packages/60/52/b3958f17315854ca781909f39392e6b40d45965820912cd610191597bd92/humming_kernels-0.1.17.tar.gz", hash = "sha256:a2565bf03b8b2ca07086db00a949c0c92727ea675e4d3037f9337192a840225a", size = 403057, upload-time = "2026-10-14T17:07:29.566Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/6d/41/288bf756d921dbe98982eeb3ec4c20e7cb5224ea6dcb164f2df3d2f68a7f/humming_kernels-0.1.2-py3-none-any.whl", hash = "sha256:f7434b0424946445ef5ad5682bcabf309d97721818ed5bdc4c6f61de3c6b9d2f", size = 160951, upload-time = "2026-05-23T16:18:06.405Z" }, + { url = "https://files.pythonhosted.org/packages/38/78/ee230d0165b47f08103fdb9a58a64b28c91e15ed7e2d3beb3920543a9f42/humming_kernels-0.1.17-cp310-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:83dda343f36a52efb617ab7b85747baabfa1c2d00dcf1078c45f92b8df7a01c6", size = 496273, upload-time = "2026-10-14T17:07:26.658Z" }, + { url = "https://files.pythonhosted.org/packages/da/fa/c4e367ca51b2611faf88ac9ee1abe5e9ee6d4d174bbabdb54481c65f7e2e/humming_kernels-0.1.17-cp310-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:e418b5482dabd71f64622893bdbb8bf27eb30379c443b92d32af48825d9ad64c", size = 504747, upload-time = "2026-10-14T17:07:28.196Z" }, ] [package.optional-dependencies] cu13 = [ - { name = "nvidia-cuda-cccl" }, - { name = "nvidia-cuda-nvcc" }, - { name = "nvidia-cuda-nvrtc" }, - { name = "nvidia-cuda-runtime" }, + { name = "nvidia-cuda-cccl", marker = "python_full_version >= '3.11'" }, + { name = "nvidia-cuda-nvcc", marker = "python_full_version >= '3.11'" }, + { name = "nvidia-cuda-nvrtc", marker = "python_full_version >= '3.11'" }, + { name = "nvidia-cuda-runtime", marker = "python_full_version >= '3.11'" }, ] [[package]] @@ -2939,11 +3686,59 @@ wheels = [ name = "idna" version = "3.15" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] sdist = { url = "https://files.pythonhosted.org/packages/82/77/7b3966d0b9d1d31a36ddf1746926a11dface89a83409bf1483f0237aa758/idna-3.15.tar.gz", hash = "sha256:ca962446ea538f7092a95e057da437618e886f4d349216d2b1e294abfdb65fdc", size = 199245, upload-time = "2026-05-12T22:45:57.011Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/d2/23/408243171aa9aaba178d3e2559159c24c1171a641aa83b67bdd3394ead8e/idna-3.15-py3-none-any.whl", hash = "sha256:048adeaf8c2d788c40fee287673ccaa74c24ffd8dcf09ffa555a2fbb59f10ac8", size = 72340, upload-time = "2026-05-12T22:45:55.733Z" }, ] +[[package]] +name = "idna" +version = "3.20" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +sdist = { url = "https://files.pythonhosted.org/packages/f5/08/8eea9d4b8302028f3abb2c0813953f7aec26d33b7a8960ed760e65ff29fa/idna-3.20.tar.gz", hash = "sha256:a7db850025b95ded1eae8a46181a1a6c56c92c96f0e2b005d9ff8dc0210cab44", size = 216463, upload-time = "2026-09-17T14:11:04.752Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/58/a2/bb081bab032533a855d44de1d56f8e8426114ff1ba5d1f07a438a0a654f8/idna-3.20-py3-none-any.whl", hash = "sha256:ab7ae7122974553370f0bdb919e1a960b2cd1bc1ef0276416d896db81c14582c", size = 69583, upload-time = "2026-09-17T14:11:03.168Z" }, +] + [[package]] name = "ijson" version = "3.5.0" @@ -3056,6 +3851,27 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/cb/b1/3846dd7f199d53cb17f49cba7e651e9ce294d8497c8c150530ed11865bb8/iniconfig-2.3.0-py3-none-any.whl", hash = "sha256:f631c04d2c48c52b84d0d0549c99ff3859c98df65b3101406327ecc7d53fbf12", size = 7484, upload-time = "2025-10-18T21:55:41.639Z" }, ] +[[package]] +name = "instanttensor" +version = "0.2.2" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/ca/a1/95d0d4cee908ab0bd8e56b44cdb19e678081c54de52d280b0b3c9f7e4c43/instanttensor-0.2.2.tar.gz", hash = "sha256:aca07c1afe4fde23dcbfb53044e2bef50dad4c0af054317acc669a1c8dc2b325", size = 613106, upload-time = "2026-10-13T15:06:59.034Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/eb/cd/06bd5595b5f58738e1f32d9c96e657955a44ce510d00b09c0d1bd156d2a9/instanttensor-0.2.2-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:559bbce844f6ce043ade31ac73eb67026da8f7c648a37cbbc86d3baf8aec408a", size = 6362824, upload-time = "2026-10-13T15:05:33.294Z" }, + { url = "https://files.pythonhosted.org/packages/23/78/e272f01c0b10549180d58c73a26ac57456854ae0c98d652c81f34ac46938/instanttensor-0.2.2-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:1ba19b3c7a9099a6e7ffe644e8bc808244ceb0b3dcefedda48243d90b89cc328", size = 6431069, upload-time = "2026-10-13T15:05:39.567Z" }, + { url = "https://files.pythonhosted.org/packages/b8/51/00cd5124d244c9dc988e294bdb4bb85d0355829d6d818a9b459b585ec452/instanttensor-0.2.2-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d43097e15922fd90e45a19308efc2ec63290e51f308f8501eaae705d7249df87", size = 6425003, upload-time = "2026-10-13T15:05:45.531Z" }, + { url = "https://files.pythonhosted.org/packages/a1/5d/b8dcff8c81bffe875344b033704fda7a0bc30bf65ff808fff0783f6b3adf/instanttensor-0.2.2-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:78ba35382f42e81d18aa9ab59485ff1d1764c846de68bc5de8b894b37ee14996", size = 6509568, upload-time = "2026-10-13T15:05:51.7Z" }, + { url = "https://files.pythonhosted.org/packages/ff/08/a28ad4c33e42f45d5617f8feb24137fcc1b7af7a7f0068c3fcf3448a0806/instanttensor-0.2.2-cp312-cp312-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:94dabc13981fe89982a807572593c4f697a3a3bae02fa13cb469960885901247", size = 6439035, upload-time = "2026-10-13T15:06:00.734Z" }, + { url = "https://files.pythonhosted.org/packages/59/12/0a4b55402ec7f45f223d4cab19c8d6d39f5242d386f4f421e30c34dd8900/instanttensor-0.2.2-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:382954f2329fc085ff298ce0205a06ce0fc8c67a704fc032613931a5ae9018d5", size = 6522862, upload-time = "2026-10-13T15:06:07.285Z" }, + { url = "https://files.pythonhosted.org/packages/cd/d7/b7a4d6f57a7b5d76377ab0660f152e4cabb3ccc4a55c36e14f7fd6ecbda1/instanttensor-0.2.2-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:fd1757e57d9561a6c28dfb18568890fd50d1f3e331d7ec247034c011f3086be9", size = 6436814, upload-time = "2026-10-13T15:06:14.375Z" }, + { url = "https://files.pythonhosted.org/packages/54/45/0ade43f9fa2ae52bcdb7c1dad7ae190919f4e1a97a2f88dc6131ac00fb37/instanttensor-0.2.2-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0474ae1f12bccbd01fabaeffab483870d9be644dc5c52d8b746ee1806dbc02fe", size = 6525208, upload-time = "2026-10-13T15:06:22.177Z" }, + { url = "https://files.pythonhosted.org/packages/56/b7/a659fc9f2e2510a127216beb1eb909fec6bacb6c404f903e3e307daf5957/instanttensor-0.2.2-cp314-cp314-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4b5ea4b516a372ef7f54e33324be10c289288f7ef57a0667b8a321293091c249", size = 6415230, upload-time = "2026-10-13T15:06:44.828Z" }, + { url = "https://files.pythonhosted.org/packages/65/2e/7bf9604ed24f24a254c20754c0ded92c416242ab1f24628d614a41f548f1/instanttensor-0.2.2-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0ccf832e544dcd13aac8da280ce70ad1b2bfede4fc00a7e7e3e6f963e3dd7eb4", size = 6520762, upload-time = "2026-10-13T15:06:56.812Z" }, +] + [[package]] name = "interegular" version = "0.3.3" @@ -3099,8 +3915,10 @@ name = "ipython" version = "9.10.1" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -3128,18 +3946,26 @@ name = "ipython" version = "9.12.0" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -3377,7 +4203,8 @@ name = "kernels" version = "0.12.3" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "huggingface-hub" }, + { name = "huggingface-hub", version = "1.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "huggingface-hub", version = "1.33.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "packaging" }, { name = "pyyaml" }, { name = "tomli", marker = "python_full_version < '3.11'" }, @@ -3624,7 +4451,8 @@ source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, - { name = "torch" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11' or extra == 'extra-7-vec-inf-sglang'" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/02/1a/0d6d0269cf7327e31a2fe80b274d5bf5001dc4462acbef26240da6ac5dfe/linear_operator-0.6.1.tar.gz", hash = "sha256:3fba49a8080d16f822a5d870f462279cd6afbcf4ed670f4511b38fad96f61831", size = 181898, upload-time = "2026-02-27T23:43:18.716Z" } wheels = [ @@ -3663,18 +4491,26 @@ name = "llguidance" version = "1.7.6" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32'", - "python_full_version >= '3.14' and sys_platform == 'emscripten'", - "python_full_version >= '3.14' and sys_platform == 'darwin'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", - "python_full_version == '3.13.*' and sys_platform == 'darwin'", - "python_full_version == '3.13.*' and sys_platform != 'darwin'", - "python_full_version == '3.12.*' and sys_platform == 'darwin'", - "python_full_version == '3.12.*' and sys_platform != 'darwin'", - "python_full_version == '3.11.*' and sys_platform == 'darwin'", - "python_full_version == '3.11.*' and sys_platform != 'darwin'", - "python_full_version < '3.11' and sys_platform == 'darwin'", - "python_full_version < '3.11' and sys_platform != 'darwin'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin'", ] sdist = { url = "https://files.pythonhosted.org/packages/da/91/6bc8bb503dc259e46d253b5424385a54fe06c38a4c7a12befe69a3c2455a/llguidance-1.7.6.tar.gz", hash = "sha256:db7febbe412ed2015501904646750071d7e00e6df7f85c4b956ad4f206fd2df7", size = 1156574, upload-time = "2026-06-03T20:13:25.316Z" } wheels = [ @@ -3732,11 +4568,11 @@ name = "lm-format-enforcer" version = "0.11.3" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "interegular" }, - { name = "packaging" }, - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "interegular", marker = "python_full_version >= '3.11'" }, + { name = "packaging", marker = "python_full_version >= '3.11'" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "pyyaml" }, + { name = "pyyaml", marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/84/d5/41cd417ba7dfdbbcfe46cebf81fb3dfd7c591b89897560ad05bb410a465d/lm_format_enforcer-0.11.3.tar.gz", hash = "sha256:e68081c108719cce284a9bcc889709b26ffb085a1945b5eba3a12cfa96d528da", size = 40258, upload-time = "2025-08-24T19:37:47.527Z" } wheels = [ @@ -3748,8 +4584,8 @@ name = "loguru" version = "0.7.3" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "colorama", marker = "sys_platform == 'win32'" }, - { name = "win32-setctime", marker = "sys_platform == 'win32'" }, + { name = "colorama", marker = "(python_full_version >= '3.11' and sys_platform == 'win32' and extra == 'extra-7-vec-inf-vllm') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "win32-setctime", marker = "(python_full_version >= '3.11' and sys_platform == 'win32' and extra == 'extra-7-vec-inf-vllm') or (sys_platform == 'win32' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/3a/05/a1dae3dffd1116099471c643b8924f5aa6524411dc6c63fdae648c4f1aca/loguru-0.7.3.tar.gz", hash = "sha256:19480589e77d47b8d85b2c827ad95d49bf31b0dcde16593892eb51dd18706eb6", size = 63559, upload-time = "2024-12-06T11:20:56.608Z" } wheels = [ @@ -4068,29 +4904,42 @@ wheels = [ [[package]] name = "mcp" -version = "1.30.0" +version = "2.3.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "anyio" }, - { name = "httpx" }, - { name = "httpx-sse" }, - { name = "jsonschema" }, - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "anyio", marker = "python_full_version >= '3.11'" }, + { name = "httpx2", marker = "python_full_version >= '3.11'" }, + { name = "jsonschema", marker = "python_full_version >= '3.11'" }, + { name = "mcp-types", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-api", marker = "python_full_version >= '3.11'" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "pydantic-settings" }, - { name = "pyjwt", extra = ["crypto"], marker = "extra == 'extra-7-vec-inf-vllm'" }, - { name = "python-multipart" }, - { name = "pywin32", marker = "sys_platform == 'win32'" }, - { name = "sse-starlette" }, - { name = "starlette" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-inspection" }, - { name = "uvicorn", marker = "sys_platform != 'emscripten'" }, + { name = "pyjwt", extra = ["crypto"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "python-multipart", marker = "python_full_version >= '3.11'" }, + { name = "pywin32", marker = "python_full_version >= '3.11' and sys_platform == 'win32'" }, + { name = "sse-starlette", marker = "python_full_version >= '3.11'" }, + { name = "starlette", marker = "python_full_version >= '3.11'" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "typing-inspection", marker = "python_full_version >= '3.11'" }, + { name = "uvicorn", marker = "python_full_version >= '3.11' and sys_platform != 'emscripten'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/9d/8d/e0d339616f4810e9051d4aba6887afab289ab1f81875fe908b606cdfd0e3/mcp-2.3.0.tar.gz", hash = "sha256:8b147a50441cf059dc88c684e0aeed3687f0aa0f39c6cde7b90330effd2b34d8", size = 4171955, upload-time = "2026-10-02T22:06:56.091Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/20/d7/4a2878f128df729db3ca04e71bfce26405cc2e0c9ebe19c592feeb4c61b9/mcp-2.3.0-py3-none-any.whl", hash = "sha256:dd0c44c089d16453e8ae31a3877a0054d7a2314caaa81f5e0541b9b1734b2377", size = 368664, upload-time = "2026-10-02T22:06:52.119Z" }, +] + +[[package]] +name = "mcp-types" +version = "2.3.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/ba/93/0142dc84a666daf8ad51a34268f34c12fd6fda4f3810c4be2504eecc8212/mcp-1.30.0.tar.gz", hash = "sha256:445414625fce5c295faa505bb11bacece661ab6f4028d57c935db57820b7a3e4", size = 680511, upload-time = "2026-09-07T14:34:15.845Z" } +sdist = { url = "https://files.pythonhosted.org/packages/9e/2d/7c251e34207f6c51000312fc8839111ac45cfe02023f90b44e7f1051dd8e/mcp_types-2.3.0.tar.gz", hash = "sha256:d1e46549edb35ee19a94940fcee6d1addd7e589ab7ea92dda83f5d84781fc362", size = 65898, upload-time = "2026-10-02T22:06:57.863Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/f5/f4/e58bc33317c92a0203664daaf00bf6f41166cc0149e5d6870a03f7cd004a/mcp-1.30.0-py3-none-any.whl", hash = "sha256:666edb5009503e1047c9d60346a756f94b261f05cc2625f23d41c728ffc484d0", size = 234581, upload-time = "2026-09-07T14:34:14.266Z" }, + { url = "https://files.pythonhosted.org/packages/ce/48/32583066b155bb5d4507e7887dc88ee23118cc4de8e2a12870c49719e2a8/mcp_types-2.3.0-py3-none-any.whl", hash = "sha256:968efdbdaedfab06adae40d378a34395f1090c5921d4be3c9cde283aaf76d91d", size = 69108, upload-time = "2026-10-02T22:06:54.099Z" }, ] [[package]] @@ -4132,30 +4981,82 @@ wheels = [ name = "mistral-common" version = "1.11.2" source = { registry = "https://pypi.org/simple" } -dependencies = [ - { name = "jsonschema" }, - { name = "numpy" }, - { name = "pillow" }, - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "pydantic-extra-types", extra = ["pycountry"], marker = "extra == 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "requests" }, - { name = "tiktoken" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform != 'darwin'", +] +dependencies = [ + { name = "jsonschema", marker = "python_full_version < '3.11'" }, + { name = "numpy", marker = "python_full_version < '3.11'" }, + { name = "pillow", marker = "python_full_version < '3.11'" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "pydantic-extra-types", extra = ["pycountry"], marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "requests", marker = "python_full_version < '3.11'" }, + { name = "tiktoken", marker = "python_full_version < '3.11'" }, + { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/c2/eb/12167a1bea9714582e5b4f539f9c019323363e314a499c72855ff0e5ad43/mistral_common-1.11.2.tar.gz", hash = "sha256:79f68fc2d1190f28637f40e053f919c8c2697e00b2aa679ddee562a95183f4ad", size = 6357845, upload-time = "2026-05-04T19:47:40.413Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/47/f0/6a5d604b972e442b9d36c117d01788feddad099e4965699e3516ee6fefc3/mistral_common-1.11.2-py3-none-any.whl", hash = "sha256:ebb42062cd705a0aa2bc69b4cde2b83d446ae58150b7e29322c90cb08fcfca6c", size = 6531968, upload-time = "2026-05-04T19:47:37.718Z" }, ] +[[package]] +name = "mistral-common" +version = "1.12.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "jsonschema", marker = "python_full_version >= '3.11'" }, + { name = "numpy", marker = "python_full_version >= '3.11'" }, + { name = "pillow", marker = "python_full_version >= '3.11'" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-sglang') or (python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic-extra-types", extra = ["pycountry"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "requests", marker = "python_full_version >= '3.11'" }, + { name = "tiktoken", marker = "python_full_version >= '3.11'" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/45/ee/e7860b0c9919a74fe43664f0bf19508df15243a27825aebc500b6c516d3b/mistral_common-1.12.0.tar.gz", hash = "sha256:d0f150926733cf422d4750ca19b47568df1f714557c9054a432b866f1f30bf52", size = 6416138, upload-time = "2026-09-22T15:17:56.43Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/c0/78/2881bae14f5d79c6d4626439e74f1bc1591d0f10a9eb7eea5a408ecdb877/mistral_common-1.12.0-py3-none-any.whl", hash = "sha256:fa4504b66c30c0201ae4578c0340c5ee2abd22151c271532f62e373b985a53cf", size = 6580267, upload-time = "2026-09-22T15:17:53.665Z" }, +] + [package.optional-dependencies] audio = [ - { name = "soundfile" }, - { name = "soxr" }, + { name = "soundfile", marker = "python_full_version >= '3.11'" }, + { name = "soxr", marker = "python_full_version >= '3.11'" }, ] image = [ - { name = "opencv-python-headless" }, + { name = "opencv-python-headless", marker = "python_full_version >= '3.11'" }, ] [[package]] @@ -4163,7 +5064,8 @@ name = "mkdocs" version = "1.6.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "click" }, + { name = "click", version = "8.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "colorama", marker = "sys_platform == 'win32'" }, { name = "ghp-import" }, { name = "jinja2" }, @@ -4278,7 +5180,7 @@ name = "ml-dtypes" version = "0.5.4" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy" }, + { name = "numpy", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/0e/4a/c27b42ed9b1c7d13d9ba8b6905dece787d6259152f2309338aed29b2447b/ml_dtypes-0.5.4.tar.gz", hash = "sha256:8ab06a50fb9bf9666dd0fe5dfb4676fa2b0ac0f31ecff72a6c3af8e22c063453", size = 692314, upload-time = "2025-11-17T22:32:31.031Z" } wheels = [ @@ -4323,14 +5225,14 @@ name = "model-hosting-container-standards" version = "0.1.14" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "fastapi" }, - { name = "httpx" }, - { name = "jmespath" }, - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "fastapi", marker = "python_full_version >= '3.11'" }, + { name = "httpx", marker = "python_full_version >= '3.11'" }, + { name = "jmespath", marker = "python_full_version >= '3.11'" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "setuptools" }, - { name = "starlette" }, - { name = "supervisor" }, + { name = "setuptools", marker = "python_full_version >= '3.11'" }, + { name = "starlette", marker = "python_full_version >= '3.11'" }, + { name = "supervisor", marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/c6/3d/cf5c6029648cb0a116f7b5c2f74aa155ab0c6dd723a1f204a6d7ff354526/model_hosting_container_standards-0.1.14.tar.gz", hash = "sha256:b6cf4c46d88ce6acd6e543a578bb88ffd55d1179a7c09c22e61ae1d8a567c564", size = 90386, upload-time = "2026-03-18T21:25:14.513Z" } wheels = [ @@ -4708,7 +5610,7 @@ name = "multiprocess" version = "0.70.19" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "dill" }, + { name = "dill", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/a2/f2/e783ac7f2aeeed14e9e12801f22529cc7e6b7ab80928d6dcce4e9f00922d/multiprocess-0.70.19.tar.gz", hash = "sha256:952021e0e6c55a4a9fe4cd787895b86e239a40e76802a789d6305398d3975897", size = 2079989, upload-time = "2026-01-19T06:47:39.744Z" } wheels = [ @@ -4820,6 +5722,57 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/28/88/4789719fbbe166d12d345b3ac66b96105f10001b16e00a9765ba29261a21/nbqa-1.9.1-py3-none-any.whl", hash = "sha256:95552d2f6c2c038136252a805aa78d85018aef922586270c3a074332737282e5", size = 35259, upload-time = "2024-11-10T12:21:56.731Z" }, ] +[[package]] +name = "nccl-extensions" +version = "0.1.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "cuda-core", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cuda-pathfinder", version = "1.8.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nccl4py", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "numpy", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "packaging", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/a9/aa/f5d3a72f1f06745e4cc06afb1aca27ba18d96c684184f2885b80c01dbfa0/nccl_extensions-0.1.0-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1739bd8d22d3118948bcabb05fda19c375b572301518d8e77e4dd622cab976f3", size = 22141378, upload-time = "2026-09-02T15:36:56.152Z" }, + { url = "https://files.pythonhosted.org/packages/a9/f0/1058d305f9bb76faaa6d3dc7bbd753b79da2f1a476b984ccad8bc5383231/nccl_extensions-0.1.0-cp310-cp310-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a76dca0d039640e9a49447d679beaa0307f5df5c1b46efd6772ce955bd7b16df", size = 22402630, upload-time = "2026-09-02T15:37:18.291Z" }, + { url = "https://files.pythonhosted.org/packages/81/70/8cb06143dc7926020f717d649d91b9302eb3346e29b8fdb13b257ef78d4b/nccl_extensions-0.1.0-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b5130d94bbc02852f1deb5ab38e944dc7e0db0f1b4769f5f3c8d652eb5776a86", size = 22293375, upload-time = "2026-09-02T15:37:43.985Z" }, + { url = "https://files.pythonhosted.org/packages/01/f8/5b61524b1b1a014ad523bbde766dfa6f9de76865bbaf74aaaf3fd2a7bf37/nccl_extensions-0.1.0-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:30d42d0218f46e394a21c5663f09ee9bb8a90caf4d69f0040e5a9eb2fef7b8ad", size = 22556341, upload-time = "2026-09-02T15:38:05.357Z" }, + { url = "https://files.pythonhosted.org/packages/24/77/a02842f3e7d6aef181733f94d313a1ff4b3db2726e2c067d7946e8f34cc3/nccl_extensions-0.1.0-cp312-cp312-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d85e203e284b50dad3846d6fb0d35560378537689f23d92a16c67f7d854025d7", size = 22282239, upload-time = "2026-09-02T15:38:26.362Z" }, + { url = "https://files.pythonhosted.org/packages/f3/3c/b043c1f8ca2766d48e30e9cc5a0d65d51aef77e3a8d18dbbaa95fd5386e2/nccl_extensions-0.1.0-cp312-cp312-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:54d46bee108f0bc18cdc35dbe17bba6452d6a38a270fa5e8ccde03267460fa87", size = 22589285, upload-time = "2026-09-02T15:38:53.213Z" }, + { url = "https://files.pythonhosted.org/packages/90/8b/0d98ca96c8625af88d834230b34115fac032ed69eb84c8bde7c38b956d16/nccl_extensions-0.1.0-cp313-cp313-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:3678bce16864064a9f8774ec69f5c023cb2cbcd91db60a7f98e7c87fd4f18a50", size = 22265222, upload-time = "2026-09-02T15:39:12.533Z" }, + { url = "https://files.pythonhosted.org/packages/5b/66/0f0a3fe84ffdf6ddfb13c01a036ea370519204820cc8c53303f479c43a09/nccl_extensions-0.1.0-cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:3328dc1cfeed0ddf37caac00be8c49b3e4f727dce0e7980f8831696165619a4f", size = 22564580, upload-time = "2026-09-02T15:39:32.862Z" }, + { url = "https://files.pythonhosted.org/packages/f0/21/ad0697074c2050fd7b6ee6f1909f1bb75019b710319baf1907d6d8a65476/nccl_extensions-0.1.0-cp314-cp314-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:de0ac4801a9ca851a13b2abe3825a8c60fe2a12c637d96b675ea564b31db9152", size = 22253849, upload-time = "2026-09-02T15:39:57.124Z" }, + { url = "https://files.pythonhosted.org/packages/d9/3d/2af7630a67b2c2d96e4ad520a3a04957e92b66dbc0c886520aa96b7b5c57/nccl_extensions-0.1.0-cp314-cp314-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:28f2513696bd03453857db32a5f6144631f874efb23b5fede4c6218e6026d6b1", size = 22531955, upload-time = "2026-09-02T15:40:20.373Z" }, + { url = "https://files.pythonhosted.org/packages/6b/8e/0a04f409518bfe3703519e494e876f3b1f363692392f8523dce23e19e2a1/nccl_extensions-0.1.0-cp314-cp314t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:11226949d4daaa9c95ca1b0de4227712291954ae50f8408d97e19af2c0f45d19", size = 22368463, upload-time = "2026-09-02T15:40:40.951Z" }, + { url = "https://files.pythonhosted.org/packages/e2/af/3a54416ac5e5ec08254195216e9708b8e91371d436e5c5459348400fdb98/nccl_extensions-0.1.0-cp314-cp314t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:647f28b3a7a7d5c5c40cc7db544ed7ff362ac091b28e806de93c6215d2af45fe", size = 22546859, upload-time = "2026-09-02T15:41:02.116Z" }, +] + +[[package]] +name = "nccl4py" +version = "0.6.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "cuda-core", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cuda-pathfinder", version = "1.8.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "numpy", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "packaging", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/ed/ea/fc8fde326861c8995d539b87add6a804ff67027e90d8c95fd6df0cd52d53/nccl4py-0.6.0-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:145294bf11d797ef4457dabd7502cbdbbd6e109d9ade9c85a880b2ad638db26a", size = 3530427, upload-time = "2026-09-23T17:12:30.109Z" }, + { url = "https://files.pythonhosted.org/packages/ff/eb/d97ef14f942f8beb0ce3caf0f7fa43da20b9875d46602e027321be7201f7/nccl4py-0.6.0-cp310-cp310-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:995cb99ab4321dd3aafc7f78d0247f3aa5017439ac3d0cd356a40bde07708a8d", size = 3533972, upload-time = "2026-09-23T17:13:07.812Z" }, + { url = "https://files.pythonhosted.org/packages/ad/74/86fc2839835b6fc60e99dd459fe3b673390159c52afb14658e74bc03734b/nccl4py-0.6.0-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:db38bdd6c1df2f915eebb75aef16e7a3b968bb1b6767e6019a1e194187fdad16", size = 3693333, upload-time = "2026-09-23T17:13:27.644Z" }, + { url = "https://files.pythonhosted.org/packages/7b/f8/f190caa74e5fdcd15fdd65d495abc81cd71148a5157503d90bce17acb140/nccl4py-0.6.0-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7b75f4943edc39522bec1d3c7c59804fa2fd87a879e5744660d723440cb62755", size = 3696883, upload-time = "2026-09-23T17:13:49.368Z" }, + { url = "https://files.pythonhosted.org/packages/e5/2d/86a96d31ad8092f86f15f45d051d599a3fa7a7044b0176f27a50ff16260c/nccl4py-0.6.0-cp312-cp312-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:17f5b6cac4a7edaeb1b626afabbf29b0d316b847738dace109fb9dd71f2c6b56", size = 3626706, upload-time = "2026-09-23T17:14:14.74Z" }, + { url = "https://files.pythonhosted.org/packages/bf/c5/bd7d2f1251215881233580f402606e5a480b60f469f7dbea01d30fb99066/nccl4py-0.6.0-cp312-cp312-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f4006cf8c12c26ed29811b80858a58f8f31e4b42ea85b6c1d23401981449ab9d", size = 3695883, upload-time = "2026-09-23T17:14:37.411Z" }, + { url = "https://files.pythonhosted.org/packages/08/61/e8a5a1cf712027eff86b810f700c20eee310f543768ab6bd29d4a6134884/nccl4py-0.6.0-cp313-cp313-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4aef7e63b503940f7584d0ab75d13dcdfe2f49f3d6dc9012ee9176cd600b7933", size = 3625631, upload-time = "2026-09-23T17:15:00.655Z" }, + { url = "https://files.pythonhosted.org/packages/e1/37/a780bc5ae5594ed481501c65be1816caf504f53e56b95d5cbf68eeca9b59/nccl4py-0.6.0-cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:35f3832265dc42e3ac6d7983203203132cc49ae2d3377cdb5dba62bb608ed7bd", size = 3688146, upload-time = "2026-09-23T17:15:24.066Z" }, + { url = "https://files.pythonhosted.org/packages/a4/c3/2a3761b21dc809a899edfa2b850aadeb0ff1cb922d636729ccdbc99036cb/nccl4py-0.6.0-cp314-cp314-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:728a82439608c993f6c4075e6e6fb576ce69bc53f1ba487cec5254cbcf208218", size = 3615346, upload-time = "2026-09-23T17:15:48.845Z" }, + { url = "https://files.pythonhosted.org/packages/9e/e9/d33270fbaedd85c2c9355848462a954342f225c30dd8a8db41b46d2fe1c8/nccl4py-0.6.0-cp314-cp314-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:71c81efdcc1902985e00823f18ed063dd11c10e726e47bee0666520be94b4ee8", size = 3643678, upload-time = "2026-09-23T17:16:11.117Z" }, + { url = "https://files.pythonhosted.org/packages/05/87/cb5f31191d2eba6bd6d14b10ee157b546d113a94373ea13b0fbd4c49309d/nccl4py-0.6.0-cp314-cp314t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:ba99a790360395d9cf1529a35600c29b6451ec834b946d4237bfc881fd214e27", size = 3674996, upload-time = "2026-09-23T17:16:35.146Z" }, + { url = "https://files.pythonhosted.org/packages/7b/3b/f125fac7d499a67edd57cff1c453e16e4c218ff22c5f1f6cb348b2ad2f7d/nccl4py-0.6.0-cp314-cp314t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:20f75e850859fdc17cfc1158268dba4f402d5a16e598c827cd07ba33b0f5bc34", size = 3578131, upload-time = "2026-09-23T17:16:54.109Z" }, +] + [[package]] name = "nest-asyncio" version = "1.6.0" @@ -4850,23 +5803,33 @@ name = "networkx" version = "3.6.1" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -4916,8 +5879,8 @@ name = "numba" version = "0.65.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "llvmlite" }, - { name = "numpy" }, + { name = "llvmlite", marker = "python_full_version >= '3.11'" }, + { name = "numpy", marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/49/61/7299643b9c18d669e04be7c5bcb64d985070d07553274817b45b049e7bfe/numba-0.65.0.tar.gz", hash = "sha256:edad0d9f6682e93624c00125a471ae4df186175d71fd604c983c377cdc03e68b", size = 2764131, upload-time = "2026-04-01T03:52:01.946Z" } wheels = [ @@ -5013,9 +5976,44 @@ wheels = [ name = "nvidia-cublas" version = "13.1.0.3" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] wheels = [ { url = "https://files.pythonhosted.org/packages/e1/a5/fce49e2ae977e0ccc084e5adafceb4f0ac0c8333cb6863501618a7277f67/nvidia_cublas-13.1.0.3-py3-none-manylinux_2_27_aarch64.whl", hash = "sha256:c86fc7f7ae36d7528288c5d88098edcb7b02c633d262e7ddbb86b0ad91be5df2", size = 542851226, upload-time = "2025-10-09T08:59:04.818Z" }, { url = "https://files.pythonhosted.org/packages/e7/44/423ac00af4dd95a5aeb27207e2c0d9b7118702149bf4704c3ddb55bb7429/nvidia_cublas-13.1.0.3-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:ee8722c1f0145ab246bccb9e452153b5e0515fd094c3678df50b2a0888b8b171", size = 423133236, upload-time = "2025-10-09T08:59:32.536Z" }, + { url = "https://files.pythonhosted.org/packages/10/f5/f50bc3f5c2bb57ab8f5b4d78bc1146b57810d42cb8fcb28cbe2e14050376/nvidia_cublas-13.1.0.3-py3-none-win_amd64.whl", hash = "sha256:2a3b94a37def342471c59fad7856caee4926809a72dd5270155d6a31b5b277be", size = 404355960, upload-time = "2025-10-09T09:07:00.987Z" }, +] + +[[package]] +name = "nvidia-cublas" +version = "13.1.1.3" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "nvidia-cuda-nvrtc", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/a7/a1/0bd24ee8c8d03adac032fd2909426a00c88f8c57961b1277ded97f91119f/nvidia_cublas-13.1.1.3-py3-none-manylinux_2_27_aarch64.whl", hash = "sha256:b7a210458267ac818974c53038fbec2e969d5c99f305ab15c72522fa9f001dd5", size = 542848918, upload-time = "2026-04-08T18:46:22.985Z" }, + { url = "https://files.pythonhosted.org/packages/3b/cd/154ca20c38269e05eff77c1464e6c1da89f50a6390b565e9d82e06bc11e1/nvidia_cublas-13.1.1.3-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:37936a16db8fe4ac1f065c2139360608a543a09275cb1a1af612e08cfa065436", size = 423138758, upload-time = "2026-04-08T18:46:58.655Z" }, + { url = "https://files.pythonhosted.org/packages/45/9e/2f562daf80eb8f7a685fb7bea4fda71f6048e4f359d6fdd1b6e70206cb2f/nvidia_cublas-13.1.1.3-py3-none-win_amd64.whl", hash = "sha256:b6cdce694e47ff6aadf0a69df1cab6628d696f5ff56e8d16af50309d855fa20f", size = 404358158, upload-time = "2026-04-08T18:47:26.987Z" }, ] [[package]] @@ -5045,6 +6043,7 @@ source = { registry = "https://pypi.org/simple" } wheels = [ { url = "https://files.pythonhosted.org/packages/2a/2a/80353b103fc20ce05ef51e928daed4b6015db4aaa9162ed0997090fe2250/nvidia_cuda_cupti-13.0.85-py3-none-manylinux_2_25_aarch64.whl", hash = "sha256:796bd679890ee55fb14a94629b698b6db54bcfd833d391d5e94017dd9d7d3151", size = 10310827, upload-time = "2025-09-04T08:26:42.012Z" }, { url = "https://files.pythonhosted.org/packages/33/6d/737d164b4837a9bbd202f5ae3078975f0525a55730fe871d8ed4e3b952b0/nvidia_cuda_cupti-13.0.85-py3-none-manylinux_2_25_x86_64.whl", hash = "sha256:4eb01c08e859bf924d222250d2e8f8b8ff6d3db4721288cf35d14252a4d933c8", size = 10715597, upload-time = "2025-09-04T08:26:51.312Z" }, + { url = "https://files.pythonhosted.org/packages/ad/df/b74b10025c1205695c5676373f2edd3e87a7202cc62ead0dfbc373b0f6ea/nvidia_cuda_cupti-13.0.85-py3-none-win_amd64.whl", hash = "sha256:683f58d301548deeefcb8f6fac1b8d907691b9d8b18eccab417f51e362102f00", size = 7736776, upload-time = "2025-09-04T08:38:08.38Z" }, ] [[package]] @@ -5052,9 +6051,9 @@ name = "nvidia-cuda-nvcc" version = "13.3.33" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "nvidia-cuda-crt" }, - { name = "nvidia-cuda-runtime" }, - { name = "nvidia-nvvm" }, + { name = "nvidia-cuda-crt", marker = "python_full_version >= '3.11'" }, + { name = "nvidia-cuda-runtime", marker = "python_full_version >= '3.11'" }, + { name = "nvidia-nvvm", marker = "python_full_version >= '3.11'" }, ] wheels = [ { url = "https://files.pythonhosted.org/packages/be/b6/bb07a3a63b5b7b55516366747892abbf3ee62d616684c40bb51e6cbfe956/nvidia_cuda_nvcc-13.3.33-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:8c348623b1434aebd234da9ec1f81022587ae4995d65c3dc8a7743245cc441f7", size = 39515074, upload-time = "2026-05-26T16:34:28.489Z" }, @@ -5086,18 +6085,69 @@ wheels = [ name = "nvidia-cudnn-cu13" version = "9.19.0.56" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] dependencies = [ - { name = "nvidia-cublas" }, + { name = "nvidia-cublas", version = "13.1.0.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and sys_platform != 'darwin') or (python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] wheels = [ { url = "https://files.pythonhosted.org/packages/f1/84/26025437c1e6b61a707442184fa0c03d083b661adf3a3eecfd6d21677740/nvidia_cudnn_cu13-9.19.0.56-py3-none-manylinux_2_27_aarch64.whl", hash = "sha256:6ed29ffaee1176c612daf442e4dd6cfeb6a0caa43ddcbeb59da94953030b1be4", size = 433781201, upload-time = "2026-02-03T20:40:53.805Z" }, { url = "https://files.pythonhosted.org/packages/a3/22/0b4b932655d17a6da1b92fa92ab12844b053bb2ac2475e179ba6f043da1e/nvidia_cudnn_cu13-9.19.0.56-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:d20e1734305e9d68889a96e3f35094d733ff1f83932ebe462753973e53a572bf", size = 366066321, upload-time = "2026-02-03T20:44:52.837Z" }, + { url = "https://files.pythonhosted.org/packages/91/a2/f020386683ee9ab2c9a9f7f79290d9b0d07f7241de54dc746af2abd188d2/nvidia_cudnn_cu13-9.19.0.56-py3-none-win_amd64.whl", hash = "sha256:40d8c375005bcb01495f8edf375230b203a411a0c05fb6dc92a3781edcb23eac", size = 350547366, upload-time = "2026-02-03T20:50:49.563Z" }, +] + +[[package]] +name = "nvidia-cudnn-cu13" +version = "9.20.0.48" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "nvidia-cublas", version = "13.1.0.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cublas", version = "13.1.1.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/56/c5/83384d846b2fd17c44bd499b36c75a45ed4f095fbbb2252294e89cea5c5c/nvidia_cudnn_cu13-9.20.0.48-py3-none-manylinux_2_27_aarch64.whl", hash = "sha256:e31454ae00094b0c55319d9d15b6fa2fc50a9e1c0f5c8c80fb75258234e731e1", size = 444574296, upload-time = "2026-03-09T19:28:27.751Z" }, + { url = "https://files.pythonhosted.org/packages/6e/5e/edb9c0ae051602c3ccaffe424256463636d639e27d7f302dde9975ef9e7a/nvidia_cudnn_cu13-9.20.0.48-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:0c45dd8eeb50b603f07995b1b300c62ffe6a1980482b82b3bcf94a4ca9d49304", size = 366173588, upload-time = "2026-03-09T19:29:34.474Z" }, + { url = "https://files.pythonhosted.org/packages/78/39/21507455b1bca8b5702a9e9fc6ce73735f216f558dac2c9ede58e4d456b8/nvidia_cudnn_cu13-9.20.0.48-py3-none-win_amd64.whl", hash = "sha256:af8139732b99c0118be65ea5aac97f0d46018f8c552889e49d2fb0c6261a4a24", size = 350712614, upload-time = "2026-03-09T19:31:11.398Z" }, ] [[package]] name = "nvidia-cudnn-frontend" version = "1.18.0" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] wheels = [ { url = "https://files.pythonhosted.org/packages/86/be/f5a1e633c524c13c0182213ab27dab42dca29a3c785be5ff74d2d185aed1/nvidia_cudnn_frontend-1.18.0-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:baa6fbc8e7c55f1c78c0374ed9a890e1cf81acaca0c92d6135d18a8e3c985244", size = 2023500, upload-time = "2026-01-27T23:31:34.747Z" }, { url = "https://files.pythonhosted.org/packages/82/a7/765a17c6a9496196c34f269d17dfb902b6c618c0261c0962511e95302e81/nvidia_cudnn_frontend-1.18.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:2e4bcca42259e358002c8867e3624a558f66cd5dff2cc6c3aafd860ef2f41730", size = 2154278, upload-time = "2026-01-27T23:06:55.784Z" }, @@ -5116,16 +6166,70 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/67/53/df2810b56d259ef96fa6beaa1381bd14c29fbe82836b409516e864c5e177/nvidia_cudnn_frontend-1.18.0-cp314-cp314-win_amd64.whl", hash = "sha256:5053b473fa74168b5fbf35934cd6187f88aa03b8447b9f2cd417332d5e5c9569", size = 1592759, upload-time = "2026-02-11T21:32:33.87Z" }, ] +[[package]] +name = "nvidia-cudnn-frontend" +version = "1.31.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "apache-tvm-ffi", version = "0.1.11", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cutlass-dsl", version = "4.8.0", source = { registry = "https://pypi.org/simple" }, extra = ["cu13"], marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/30/c5/fb5ae029490072b139b50dbcfdf26f668e45f274e9f12596a1792dbd5499/nvidia_cudnn_frontend-1.31.0-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4bb7cad9c6e41003f149717f072c70139ed73243172bd8f64f12222e8b1556f9", size = 9963267, upload-time = "2026-10-08T18:03:30.491Z" }, + { url = "https://files.pythonhosted.org/packages/83/aa/15baf4db6e5c6f547aeafff8e158827286f7bd2347d59e59d8b1286ea791/nvidia_cudnn_frontend-1.31.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f7ae50ce35cdca2a995f2a0765ba95b60ca4a28c157624ad099df4b69a8439e4", size = 10146384, upload-time = "2026-10-08T18:03:52.798Z" }, + { url = "https://files.pythonhosted.org/packages/1e/51/e8951d811bb16ef33550aafb42e4a82d4ee3d88e1e5526a2331f52f89604/nvidia_cudnn_frontend-1.31.0-cp310-cp310-win_amd64.whl", hash = "sha256:cd578e982f76e627dfa5ed8900237a53696e9dacc97447fbd327f45d8211699f", size = 9365142, upload-time = "2026-10-08T18:04:12.746Z" }, + { url = "https://files.pythonhosted.org/packages/64/19/10c2ccd549b064ac54885bbdbbb9ceb94e329c1e26716f4529331619429e/nvidia_cudnn_frontend-1.31.0-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5d1ae02680c2e75e32f12a7c1a189082aaac6864451d64f4c4c2c9ba0616504e", size = 9966093, upload-time = "2026-10-08T18:04:31.519Z" }, + { url = "https://files.pythonhosted.org/packages/b7/1a/dd78de8a3f2711de7e282f0c2b2365dbdbfb079047d12df15ec9aa2a8d3b/nvidia_cudnn_frontend-1.31.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:be3831ed007147d4720f7821501aeb448456074615326363ec2ea3ad6e6c45d3", size = 10148839, upload-time = "2026-10-08T18:04:54.257Z" }, + { url = "https://files.pythonhosted.org/packages/21/b2/62b8f096df9faf092d42b8f86c355735395026f4a080a762cd861a5ee85d/nvidia_cudnn_frontend-1.31.0-cp311-cp311-win_amd64.whl", hash = "sha256:f9a04bd0b5fc0ca5bdff4a0a423b443c8edc879aa2ae954ce8c328eb6a5c9fc4", size = 9366380, upload-time = "2026-10-08T18:05:26.995Z" }, + { url = "https://files.pythonhosted.org/packages/bf/4b/ba084b26c9709b05446642413aa4bbb1056fe00af7f99e356a3cf24774d6/nvidia_cudnn_frontend-1.31.0-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:47d1a5432f59a2a7d0f903727c0d244e1413a20686b98d8e3890919c5d82a508", size = 9970357, upload-time = "2026-10-08T18:05:50.547Z" }, + { url = "https://files.pythonhosted.org/packages/cb/24/8ec9de1fd9ea95e1c1cb9e6e2d3a5547aa9c5ece1136d650ffce1bb5d8c5/nvidia_cudnn_frontend-1.31.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:3171f102254136418f24b4470dfa5419f34b03d9d792a4b773cf65c7a37e6bf8", size = 10159344, upload-time = "2026-10-08T18:06:15.766Z" }, + { url = "https://files.pythonhosted.org/packages/4f/b6/caa5b41361034e7884b465afa2214137ec325be8ea099553f6e597f75af1/nvidia_cudnn_frontend-1.31.0-cp312-cp312-win_amd64.whl", hash = "sha256:786c7dc590d436338c55b74e3204fc2adb4739acafb1a5e46ce032282e601e2e", size = 9373760, upload-time = "2026-10-08T18:06:35.091Z" }, + { url = "https://files.pythonhosted.org/packages/d8/fe/b4263f0f05befaa9ebc760ae94229363ac66e4909e622d673691a0eb35b3/nvidia_cudnn_frontend-1.31.0-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b01d93e6ed709c349aa33a3db32075910ffccccd40b017aa293d93a6172c64b6", size = 9970347, upload-time = "2026-10-08T18:06:59.4Z" }, + { url = "https://files.pythonhosted.org/packages/59/44/8bccef95102ed1b536b05aaf9a4b6d1c3bd51cbe98a31a9718efa5ce15ed/nvidia_cudnn_frontend-1.31.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:8095ca4f3bc86e44e5da16584e951a0c33030962d9f647baa20cc1a848188b68", size = 10159077, upload-time = "2026-10-08T18:07:20.362Z" }, + { url = "https://files.pythonhosted.org/packages/89/d2/92a862203420b6d00dd3634b24ebebf20341be9ee2ee7e632f74f0d4c4f9/nvidia_cudnn_frontend-1.31.0-cp313-cp313-win_amd64.whl", hash = "sha256:a30b02bd92415a82c448ad988b7c5cbb96bd9654d88b4137a7ac37db5aa7ac9a", size = 9373599, upload-time = "2026-10-08T18:07:41.978Z" }, + { url = "https://files.pythonhosted.org/packages/55/80/dfc27b7a6827b58c9b4a5dfeb424ef07549bcd82390b95f5eb222f008b12/nvidia_cudnn_frontend-1.31.0-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:0d5f26cc1f9a2869f7ec05b2155e1e04bcb271a4b7272a310fddad7f34e8347a", size = 9973271, upload-time = "2026-10-08T18:08:06.42Z" }, + { url = "https://files.pythonhosted.org/packages/35/f5/890417a7317704edcfc260ab3186af15417e30d7f27942ca9cba1cf9f64c/nvidia_cudnn_frontend-1.31.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:9f9ca9fd86103057bb80d40f848090a3dd2d39758596b0555f024b90f062009b", size = 10160864, upload-time = "2026-10-08T18:08:28.143Z" }, + { url = "https://files.pythonhosted.org/packages/45/f5/9c441440d97f9964199f3be4ac3d7e7ffba6f7c67762eb846ced32e2d46c/nvidia_cudnn_frontend-1.31.0-cp314-cp314-win_amd64.whl", hash = "sha256:1168c10f7dabbaa261b34794ef51a7ec5f5d1c6d73156e765045ca62bc6d7bfd", size = 9371312, upload-time = "2026-10-08T18:08:48.566Z" }, + { url = "https://files.pythonhosted.org/packages/04/7a/998103de44dadc6383afec7fe241e8d300b953d8e2f64c46c891b8943ca9/nvidia_cudnn_frontend-1.31.0-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:11286b8798f417b43532a676cf881aa11181ba9b4b726ab048cd3286b53f7d31", size = 9976196, upload-time = "2026-10-08T18:09:11.634Z" }, + { url = "https://files.pythonhosted.org/packages/6b/66/a2f2aadd6d0c5e8afa333c4fa5daf5e3007c484e228be5c16c196ff41b2f/nvidia_cudnn_frontend-1.31.0-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:640205cc0eab9df478e3c46a068bf0fe407ef185a2614e710c0d4f9f64c4efc0", size = 10154698, upload-time = "2026-10-08T18:09:34.655Z" }, + { url = "https://files.pythonhosted.org/packages/96/6b/55da0c1e8e27056fca18396618bc60660c9238e9b8c64f3a9685de874288/nvidia_cudnn_frontend-1.31.0-cp314-cp314t-win_amd64.whl", hash = "sha256:dc2cd889dc72e16768b773f15f53af1fac7646173c4a1e57f41c673dd45c52b0", size = 9408214, upload-time = "2026-10-08T18:10:00.281Z" }, +] + [[package]] name = "nvidia-cufft" version = "12.0.0.61" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "nvidia-nvjitlink" }, + { name = "nvidia-nvjitlink", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten') or (python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'win32') or (python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (python_full_version < '3.11' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32') or (platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32') or (sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-vllm') or (sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-vllm') or (sys_platform == 'win32' and extra != 'extra-7-vec-inf-vllm') or (extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] wheels = [ { url = "https://files.pythonhosted.org/packages/8b/ae/f417a75c0259e85c1d2f83ca4e960289a5f814ed0cea74d18c353d3e989d/nvidia_cufft-12.0.0.61-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:2708c852ef8cd89d1d2068bdbece0aa188813a0c934db3779b9b1faa8442e5f5", size = 214053554, upload-time = "2025-09-04T08:31:38.196Z" }, { url = "https://files.pythonhosted.org/packages/a8/2f/7b57e29836ea8714f81e9898409196f47d772d5ddedddf1592eadb8ab743/nvidia_cufft-12.0.0.61-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:6c44f692dce8fd5ffd3e3df134b6cdb9c2f72d99cf40b62c32dde45eea9ddad3", size = 214085489, upload-time = "2025-09-04T08:31:56.044Z" }, + { url = "https://files.pythonhosted.org/packages/85/b2/f8af21a2ed1beed337a6a02c5a28aeb85441f4d578ec3d529543c775ea4b/nvidia_cufft-12.0.0.61-py3-none-win_amd64.whl", hash = "sha256:2abce5b39d2f5ae12730fb7e5db6696533e36c26e2d3e8fd1750bdd2853364eb", size = 213342123, upload-time = "2025-09-04T08:40:51.145Z" }, ] [[package]] @@ -5144,6 +6248,7 @@ source = { registry = "https://pypi.org/simple" } wheels = [ { url = "https://files.pythonhosted.org/packages/1e/72/7c2ae24fb6b63a32e6ae5d241cc65263ea18d08802aaae087d9f013335a2/nvidia_curand-10.4.0.35-py3-none-manylinux_2_27_aarch64.whl", hash = "sha256:133df5a7509c3e292aaa2b477afd0194f06ce4ea24d714d616ff36439cee349a", size = 61962106, upload-time = "2025-08-04T10:21:41.128Z" }, { url = "https://files.pythonhosted.org/packages/a5/9f/be0a41ca4a4917abf5cb9ae0daff1a6060cc5de950aec0396de9f3b52bc5/nvidia_curand-10.4.0.35-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:1aee33a5da6e1db083fe2b90082def8915f30f3248d5896bcec36a579d941bfc", size = 59544258, upload-time = "2025-08-04T10:22:03.992Z" }, + { url = "https://files.pythonhosted.org/packages/99/27/72103153b1ffc00e09fdc40ac970235343dcd1ea8bd762e84d2d73219ffa/nvidia_curand-10.4.0.35-py3-none-win_amd64.whl", hash = "sha256:65b1710aa6961d326b411e314b374290904c5ddf41dc3f766ebc3f1d7d4ca69f", size = 55242481, upload-time = "2025-08-04T10:30:41.831Z" }, ] [[package]] @@ -5151,13 +6256,15 @@ name = "nvidia-cusolver" version = "12.0.4.66" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "nvidia-cublas" }, - { name = "nvidia-cusparse" }, - { name = "nvidia-nvjitlink" }, + { name = "nvidia-cublas", version = "13.1.0.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and sys_platform != 'darwin') or (python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cublas", version = "13.1.1.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cusparse", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten') or (python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'win32') or (python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (python_full_version < '3.11' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32') or (platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32') or (sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-vllm') or (sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-vllm') or (sys_platform == 'win32' and extra != 'extra-7-vec-inf-vllm') or (extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-nvjitlink", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten') or (python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'win32') or (python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (python_full_version < '3.11' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32') or (platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32') or (sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-vllm') or (sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-vllm') or (sys_platform == 'win32' and extra != 'extra-7-vec-inf-vllm') or (extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] wheels = [ { url = "https://files.pythonhosted.org/packages/c8/c3/b30c9e935fc01e3da443ec0116ed1b2a009bb867f5324d3f2d7e533e776b/nvidia_cusolver-12.0.4.66-py3-none-manylinux_2_27_aarch64.whl", hash = "sha256:02c2457eaa9e39de20f880f4bd8820e6a1cfb9f9a34f820eb12a155aa5bc92d2", size = 223467760, upload-time = "2025-09-04T08:33:04.222Z" }, { url = "https://files.pythonhosted.org/packages/5f/67/cba3777620cdacb99102da4042883709c41c709f4b6323c10781a9c3aa34/nvidia_cusolver-12.0.4.66-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:0a759da5dea5c0ea10fd307de75cdeb59e7ea4fcb8add0924859b944babf1112", size = 200941980, upload-time = "2025-09-04T08:33:22.767Z" }, + { url = "https://files.pythonhosted.org/packages/99/ef/332a0101260ca78a1daef046bf0b06199e8ed4dac1d2aa698289c358169c/nvidia_cusolver-12.0.4.66-py3-none-win_amd64.whl", hash = "sha256:16515bd33a8e76bb54d024cfa068fa68d30e80fc34b9e1090813ea9362e0cb65", size = 193551444, upload-time = "2025-09-04T08:41:46.813Z" }, ] [[package]] @@ -5165,20 +6272,53 @@ name = "nvidia-cusparse" version = "12.6.3.3" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "nvidia-nvjitlink" }, + { name = "nvidia-nvjitlink", marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten') or (python_full_version >= '3.11' and python_full_version < '3.14' and platform_machine != 's390x' and sys_platform == 'win32') or (python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (python_full_version < '3.11' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32') or (platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32') or (sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-vllm') or (sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-vllm') or (sys_platform == 'win32' and extra != 'extra-7-vec-inf-vllm') or (extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] wheels = [ { url = "https://files.pythonhosted.org/packages/f8/94/5c26f33738ae35276672f12615a64bd008ed5be6d1ebcb23579285d960a9/nvidia_cusparse-12.6.3.3-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:80bcc4662f23f1054ee334a15c72b8940402975e0eab63178fc7e670aa59472c", size = 162155568, upload-time = "2025-09-04T08:33:42.864Z" }, { url = "https://files.pythonhosted.org/packages/fa/18/623c77619c31d62efd55302939756966f3ecc8d724a14dab2b75f1508850/nvidia_cusparse-12.6.3.3-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:2b3c89c88d01ee0e477cb7f82ef60a11a4bcd57b6b87c33f789350b59759360b", size = 145942937, upload-time = "2025-09-04T08:33:58.029Z" }, + { url = "https://files.pythonhosted.org/packages/02/b0/b043d6f3480f102f885cf87fc3ffd3edcb5e23b855025a50e2ef4d059185/nvidia_cusparse-12.6.3.3-py3-none-win_amd64.whl", hash = "sha256:cbcf42feb737bd7ec15b4c0a63e62351886bd3f975027b8815d7f720a2b5ea79", size = 143783033, upload-time = "2025-09-04T08:42:12.391Z" }, ] [[package]] name = "nvidia-cusparselt-cu13" version = "0.8.0" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] wheels = [ { url = "https://files.pythonhosted.org/packages/46/10/8dcd1175260706a2fc92a16a52e306b71d4c1ea0b0cc4a9484183399818a/nvidia_cusparselt_cu13-0.8.0-py3-none-manylinux2014_aarch64.whl", hash = "sha256:400c6ed1cf6780fc6efedd64ec9f1345871767e6a1a0a552a1ea0578117ea77c", size = 220791277, upload-time = "2025-08-13T19:22:40.982Z" }, { url = "https://files.pythonhosted.org/packages/fd/53/43b0d71f4e702fa9733f8b4571fdca50a8813f1e450b656c239beff12315/nvidia_cusparselt_cu13-0.8.0-py3-none-manylinux2014_x86_64.whl", hash = "sha256:25e30a8a7323935d4ad0340b95a0b69926eee755767e8e0b1cf8dd85b197d3fd", size = 169884119, upload-time = "2025-08-13T19:23:41.967Z" }, + { url = "https://files.pythonhosted.org/packages/57/de/8f0578928b9b1246d7b1324db0528e6b9f9fb54496a49f40bf71f09f1a27/nvidia_cusparselt_cu13-0.8.0-py3-none-win_amd64.whl", hash = "sha256:e80212ed7b1afc97102fbb2b5c82487aa73f6a0edfa6d26c5a152593e520bb8f", size = 156459710, upload-time = "2025-08-13T19:24:18.043Z" }, +] + +[[package]] +name = "nvidia-cusparselt-cu13" +version = "0.8.1" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/46/e1/cdc1797eadf82d3a9a575a19b33fdc871a97edbec42c00b5b5e914f4aff4/nvidia_cusparselt_cu13-0.8.1-py3-none-manylinux2014_aarch64.whl", hash = "sha256:4dca476c50bf4780d46cd0bfbd82e2bc10a08e4fef7950917ce8d7578d22a23f", size = 221051344, upload-time = "2025-09-05T18:49:51.289Z" }, + { url = "https://files.pythonhosted.org/packages/34/7d/2661f2fb3ac4302f3a246f5fc030213ac60c1fe0bce84f9783dbd831dbb7/nvidia_cusparselt_cu13-0.8.1-py3-none-manylinux2014_x86_64.whl", hash = "sha256:786ce87568c303fadb5afcc7102d454cd3040d75f6f8626f5db460d1871f4dd0", size = 170148586, upload-time = "2025-09-05T18:50:50.248Z" }, + { url = "https://files.pythonhosted.org/packages/31/83/f3647ce26916c94a6ca4ff1810623e2c405cff2dea6e78d29516b2514df9/nvidia_cusparselt_cu13-0.8.1-py3-none-win_amd64.whl", hash = "sha256:dccbd362f91a7b9024d1f55ee9f548ac065027ff15d8c8b0db889ab3a8f31215", size = 156885108, upload-time = "2025-09-05T18:51:35.958Z" }, ] [[package]] @@ -5214,32 +6354,56 @@ name = "nvidia-cutlass-dsl" version = "4.5.2" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", ] dependencies = [ - { name = "nvidia-cutlass-dsl-libs-base", version = "4.5.2", source = { registry = "https://pypi.org/simple" }, marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, + { name = "nvidia-cutlass-dsl-libs-base", version = "4.5.2", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] wheels = [ { url = "https://files.pythonhosted.org/packages/f0/15/575d7df4fe2f3406f1cfc68be72aeff2834f8a696daf1cd5bee8017e4507/nvidia_cutlass_dsl-4.5.2-py3-none-any.whl", hash = "sha256:68ed1b63ca74aae87955012da9dfd7fdaae471329d0028b229b841c7192ccf52", size = 10179, upload-time = "2026-05-25T03:38:56.364Z" }, ] +[[package]] +name = "nvidia-cutlass-dsl" +version = "4.8.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "nvidia-cutlass-dsl-libs-base", version = "4.8.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cutlass-dsl-libs-cu12", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/e4/f4/e96f452f7975dd2fd4d2a4af88c63205431dad7efbe63aff6e3a4b1ada17/nvidia_cutlass_dsl-4.8.0-py3-none-any.whl", hash = "sha256:24f996e0e5fa88c8b417f4893e615af135bdfdfdbdbed7e239faf36a88df711b", size = 10493, upload-time = "2026-09-21T03:44:08.278Z" }, +] + [package.optional-dependencies] cu13 = [ - { name = "nvidia-cutlass-dsl-libs-cu13", version = "4.5.2", source = { registry = "https://pypi.org/simple" } }, + { name = "nvidia-cutlass-dsl-libs-cu13", version = "4.8.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] [[package]] @@ -5282,26 +6446,13 @@ name = "nvidia-cutlass-dsl-libs-base" version = "4.5.2" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", ] dependencies = [ - { name = "cuda-python", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "numpy", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cuda-python", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "numpy", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] wheels = [ @@ -5319,6 +6470,108 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/15/84/08d695d2e0fa95891a2e5abd978f359d50125e4d1f056e54697d465fccc3/nvidia_cutlass_dsl_libs_base-4.5.2-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:abab8a0d2f3f5661533c366df78f973052b86a3b52b868d997a95dce5aa8f17b", size = 74514399, upload-time = "2026-05-25T03:50:20.841Z" }, ] +[[package]] +name = "nvidia-cutlass-dsl-libs-base" +version = "4.8.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "cuda-bindings", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cutlass-dsl-libs-core", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/6b/37/3c8eca872fbc747aa1e5da701f7a7e2da18ab9f18e44e4fec445b4b17f31/nvidia_cutlass_dsl_libs_base-4.8.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:f6a30ced0fbe961a839603561a03f9fa0b00834b4678301aa8da420774a5cead", size = 3551700, upload-time = "2026-09-21T03:20:53.465Z" }, + { url = "https://files.pythonhosted.org/packages/ca/85/75cb04902034d2bb61508b726a76fe1f31d38852450c694de624cc31d5d3/nvidia_cutlass_dsl_libs_base-4.8.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:4d163033400ac9cdecd2b6bce8063402076c069e8d635dbc4c48d05d072f8f28", size = 3050149, upload-time = "2026-09-21T03:21:17.174Z" }, + { url = "https://files.pythonhosted.org/packages/10/b0/ebad428375be749d8aa80747626992e9e3e4a4b28809244d68c4a0579ab6/nvidia_cutlass_dsl_libs_base-4.8.0-cp310-cp310-win_amd64.whl", hash = "sha256:76d0b57505907bd9f718d7f2b15ba38121b04e00661e66dfb9c2cde473d9373d", size = 553975, upload-time = "2026-09-21T03:21:40.29Z" }, + { url = "https://files.pythonhosted.org/packages/ba/03/cb09bad2c3ff82e3f42272b82e2953d12160f9457b5244427c313117374d/nvidia_cutlass_dsl_libs_base-4.8.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:04de038e0af658b644421c4df4c78f161a1d3c2cf9e5bcdff63b24c727dcd95c", size = 3553390, upload-time = "2026-09-21T03:21:59.359Z" }, + { url = "https://files.pythonhosted.org/packages/9e/ac/81f4d08ca7f15f42cd1abac37b909ed0ecac1a56375bb88cc2ff03254360/nvidia_cutlass_dsl_libs_base-4.8.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:ce3aec4c37b68880e85a7f3f763f8fe4d0503f49516cc855c80012e0fadf92e0", size = 3051828, upload-time = "2026-09-21T03:22:32.175Z" }, + { url = "https://files.pythonhosted.org/packages/af/2d/5bfae6e845e17942efc14ea2bb86589dd6825d679f26aa24dc7aa960d199/nvidia_cutlass_dsl_libs_base-4.8.0-cp311-cp311-win_amd64.whl", hash = "sha256:fa0c5017d8af462fa40dbb01d6d89defa013ac0508ef2e2339227b83b109f0ca", size = 553975, upload-time = "2026-09-21T03:22:54.857Z" }, + { url = "https://files.pythonhosted.org/packages/21/59/99fa03da9ace3048379601cf090a629c8d7df3ed33a4fa284848080e342d/nvidia_cutlass_dsl_libs_base-4.8.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:3affcaefb59a976ff5195dd2bff4c13594233cb4941540b74e372b63e497d3f2", size = 3553268, upload-time = "2026-09-21T03:23:16.521Z" }, + { url = "https://files.pythonhosted.org/packages/12/1c/2f7a13626af1edfefb5ad69234d47e3b212064a42ad2e6db59351b5ae002/nvidia_cutlass_dsl_libs_base-4.8.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:5bdc80989b0fbbac3d58960057e74e2ce8676fafa76b852d9422b9c4b9b74bdd", size = 3050530, upload-time = "2026-09-21T03:23:37.86Z" }, + { url = "https://files.pythonhosted.org/packages/70/fc/b17e40e75a5b120c37a67ccc41f0b596da0e0b0af90112beda4321b5a3be/nvidia_cutlass_dsl_libs_base-4.8.0-cp312-cp312-win_amd64.whl", hash = "sha256:f3ae1fc18c1353e329dc481a4010e7f738746745a20d6badea2e6035d1493c9a", size = 553975, upload-time = "2026-09-21T03:24:01.373Z" }, + { url = "https://files.pythonhosted.org/packages/bf/26/42c965639d659e2092087a176579ac3682da3007f6e1f8cd70a7d4be1615/nvidia_cutlass_dsl_libs_base-4.8.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:4d26dd2848b2fa03954151cfd6da92b79e2f4175d87f4d661eedc77c36b7a054", size = 3553164, upload-time = "2026-09-21T03:24:22.327Z" }, + { url = "https://files.pythonhosted.org/packages/74/a0/876b645cfa98137a1db319b549cc36b416d006ac352eaba5f2b2e292cf65/nvidia_cutlass_dsl_libs_base-4.8.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:5e36773748f5a8061afe999264f1c3dccc5ff18ec9548a9c769d8e72632c8b00", size = 3050412, upload-time = "2026-09-21T03:24:53.919Z" }, + { url = "https://files.pythonhosted.org/packages/61/dd/183be799302275a7e978438b58bbb61af2e1748c60ffeb407410d3604394/nvidia_cutlass_dsl_libs_base-4.8.0-cp313-cp313-win_amd64.whl", hash = "sha256:49bdb4ac523fb8e618c485c42c451ccc99a2664a6a6996ac64627fcc35a6948c", size = 553975, upload-time = "2026-09-21T03:25:15.949Z" }, + { url = "https://files.pythonhosted.org/packages/b1/20/5720c14f147559b5c6eea59971543e4ef81b3861da16d5618051f638c008/nvidia_cutlass_dsl_libs_base-4.8.0-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:752f1715d50a4af4704aa2f199986b99653b40dfb574dd0c6aa5cf264dddc800", size = 3553854, upload-time = "2026-09-21T03:25:41.565Z" }, + { url = "https://files.pythonhosted.org/packages/d1/d2/22980e2fded6fb7232b027fe1afb1a93a4b999cba293deeaf188f007e48e/nvidia_cutlass_dsl_libs_base-4.8.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:728abf1bc065fa950ad4f4506928300d7e32f3148312b93f83f3f2e8f756329a", size = 3050483, upload-time = "2026-09-21T03:26:03.194Z" }, + { url = "https://files.pythonhosted.org/packages/1a/9f/f9402062be3f2f8f6194b9b2a0b27a7ef8ce42487236b267868c87b314d8/nvidia_cutlass_dsl_libs_base-4.8.0-cp314-cp314-win_amd64.whl", hash = "sha256:25f8b98c4d42f3a41c131556616827d49c619a7a79e4884c3e36d15048cc9062", size = 554030, upload-time = "2026-09-21T03:26:22.576Z" }, + { url = "https://files.pythonhosted.org/packages/76/b4/79d87535b14f923b14cf6f45557383f7ec70aa0b02c50b1552e31320a506/nvidia_cutlass_dsl_libs_base-4.8.0-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:c137f5b1df0eba5c3bd64f49a2b72ac1748a371c7a75bd461ce4b8602fcced25", size = 3560846, upload-time = "2026-09-21T03:26:34.838Z" }, + { url = "https://files.pythonhosted.org/packages/e8/f2/f1d639c23bf4cb6f1ea483431e6a0ab6aa46bbd6e3b6a17a11ae87a6f713/nvidia_cutlass_dsl_libs_base-4.8.0-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:0c9759725b5e3967606bcaa8f28d7218729fed5f90cbb73fdac96802802f1408", size = 3063843, upload-time = "2026-09-21T03:26:58.966Z" }, + { url = "https://files.pythonhosted.org/packages/d9/8a/d57c812b47551eae2c99d40a8b86483083875fe96cfc21b1593c3f1df1fa/nvidia_cutlass_dsl_libs_base-4.8.0-cp314-cp314t-win_amd64.whl", hash = "sha256:8701b91be993b621d892796fa252aa9b63cb75fb5a7fe894536fe132b57d554e", size = 554033, upload-time = "2026-09-21T03:27:20.891Z" }, +] + +[[package]] +name = "nvidia-cutlass-dsl-libs-core" +version = "4.8.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "cuda-bindings", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/85/66/d05f0ae372b80fc757e59d5211d93483c6ff721f574cb68cad84ceb87157/nvidia_cutlass_dsl_libs_core-4.8.0-py3-none-any.whl", hash = "sha256:b32716d2ed10c185cdbbb6b5c6fca26a3113a60e6e2fe4e2f0cb15966422b621", size = 1634056, upload-time = "2026-09-21T03:20:12.741Z" }, +] + +[[package]] +name = "nvidia-cutlass-dsl-libs-cu12" +version = "4.8.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "cuda-bindings", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cutlass-dsl-libs-base", version = "4.8.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/56/56/979a2b78206790fdcc7b6051f4aa08d830c7fd6f513a16ae3cf1ba141514/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:7f4d24b96099ae7ba196a177f8893feca8e9cee0ae34172f5c4346597cd043ec", size = 90662081, upload-time = "2026-09-21T03:28:14.221Z" }, + { url = "https://files.pythonhosted.org/packages/9c/53/05531d0d3c42a25992b585c5d3dc250481a4affe25bb4409868fa23ed503/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:c75cf5efe6776fd5b34ea326d97ad07b78007e58c920fa7cfeec03d3fa5aeeac", size = 91697553, upload-time = "2026-09-21T03:28:35.823Z" }, + { url = "https://files.pythonhosted.org/packages/64/fd/0ba5b669a61e3bdc80e641a9a70177831dd2de37becfcd4238293f136f6d/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp310-cp310-win_amd64.whl", hash = "sha256:3ddb4b003c0229f2434235a5dfff977813253d0258b61f2b070dd0c88e800aaf", size = 74234521, upload-time = "2026-09-21T03:28:57.525Z" }, + { url = "https://files.pythonhosted.org/packages/9c/0c/b81c27b5f0304ad3ad4c8d195aff70e5270ea57b90837871a89150d5fd8e/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:6f88ae52d6e2aedb4a34fda32c974aa3543064cfeed3fa7e96123c32402ff2ca", size = 90663760, upload-time = "2026-09-21T03:29:21.794Z" }, + { url = "https://files.pythonhosted.org/packages/6e/6f/0a6f2988575a3f7b78c88a67fb4fffcf32e1eee31f1c5c76db0f1674f127/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:5dac7aef97fc51df882b68620e1c430e7b4362d0aca17d220727da7681f061df", size = 91698074, upload-time = "2026-09-21T03:29:42.607Z" }, + { url = "https://files.pythonhosted.org/packages/78/e8/234f1cf3c32fcbbb3dff0617072d83bf83f5cab8a58259a529a00a24b547/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp311-cp311-win_amd64.whl", hash = "sha256:b0d6bac5f04dfad8fb6c8966b3dd1f1f6d172adcc1a48b86c9a47c8464c39e01", size = 74233855, upload-time = "2026-09-21T03:30:04.563Z" }, + { url = "https://files.pythonhosted.org/packages/b9/10/6894d25cc9011416f82d33eee70ad458c5f6fd50c7e7e2f75d85b708f9a5/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:32766847abe3bef711eb6052ab78c0f24c090edd1f83322f6e74531d399e8117", size = 90663297, upload-time = "2026-09-21T03:30:26.937Z" }, + { url = "https://files.pythonhosted.org/packages/13/a0/fdde64cc231f52a716107bf1a1246db9dde8cb11b28278b8d9c05f7c87bd/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:2374cfdff635ceaf7e2c5bd17b4beb410830420a4fd65b12b7f24b5d5cbb9d4c", size = 91698149, upload-time = "2026-09-21T03:30:50.832Z" }, + { url = "https://files.pythonhosted.org/packages/72/57/c58304e2ff19ce3afd33076d6dfa6639a5023601f6fc11f7f16d6ea0eef6/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp312-cp312-win_amd64.whl", hash = "sha256:d1d942de8ea20dea9ce86fdc5c4dab6c572bab81c9f708f7e8ca54cfd78bddc4", size = 74239275, upload-time = "2026-09-21T03:31:21.136Z" }, + { url = "https://files.pythonhosted.org/packages/63/89/5cb06d4a4a59c2a5f3a8183e2a31c90b5ba679e3d53e9889ce4311580405/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:08af2b89f33081df69b64be1cad133f9f767274b33669882cbee6a5fa35c5fcc", size = 90660248, upload-time = "2026-09-21T03:31:41.285Z" }, + { url = "https://files.pythonhosted.org/packages/aa/7a/fc227e590fb9aff5d542e970011a55bb9c6a4eb3165fd64e8f4b2de631a5/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:a06ea6a42c921ec66b190f14849e152a95a45f28819214488b9b0aa34168da33", size = 91698043, upload-time = "2026-09-21T03:32:04.819Z" }, + { url = "https://files.pythonhosted.org/packages/bc/87/d84548297aa98583a483e8cca5b7db9379d0e0fb4f647472c5148e8e14da/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp313-cp313-win_amd64.whl", hash = "sha256:4963300d4242dd5ee122be0367008c57a9bb9fb11fa4d9fc8e0befb491dccecd", size = 74239680, upload-time = "2026-09-21T03:32:26.019Z" }, + { url = "https://files.pythonhosted.org/packages/71/7c/1d74bb455a41cfff0ff70b51a4a4898a8a96b991411117e7f4fbcd218d15/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:41f9d49087f18200c4516e00abade67f542795f363c7f195bb53404b81a866f8", size = 90664033, upload-time = "2026-09-21T03:32:49.163Z" }, + { url = "https://files.pythonhosted.org/packages/6f/a6/87c26f561a784631ef26a606549c26b72ca13d40886462ee8b89abceb072/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:889918c32fe6cad047ae718343fedb374a9c4275903b4ab2d8f42e811a6a0dac", size = 91698171, upload-time = "2026-09-21T03:33:11.342Z" }, + { url = "https://files.pythonhosted.org/packages/23/df/47032477e611232dc10085974139fd4a72e0c5af3fb2c8d34f02626fedc9/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp314-cp314-win_amd64.whl", hash = "sha256:74e5f9d291eb689422695e36c8c85ae65fca05c743e60d8bce392bdb9dc4f68e", size = 76219624, upload-time = "2026-09-21T03:33:33.831Z" }, + { url = "https://files.pythonhosted.org/packages/e3/17/13ccd372d4f624447e93c7a46a32a93c7df774e401d14a8d0684ef1ccf7a/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:0746ff56576d63abf616d242fd1c3884713012448461d0f2bbf25bc397b410a0", size = 90672972, upload-time = "2026-09-21T03:33:56.587Z" }, + { url = "https://files.pythonhosted.org/packages/8e/21/242fbcd98c5355589c878fd41bb3c94cb1ed62bd5e67a63f6fae93375990/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:54125669fc51e70ce97e85554ae089cc745c4b2b8d740750b315e5777304ea6c", size = 91698894, upload-time = "2026-09-21T03:34:27.435Z" }, + { url = "https://files.pythonhosted.org/packages/d3/7f/64dfbf8acc280f31719c0b8d8638332f1616c0d8ac42053b2e0e43fe2c4b/nvidia_cutlass_dsl_libs_cu12-4.8.0-cp314-cp314t-win_amd64.whl", hash = "sha256:bad2b4cab99ebf4211d305f77254b5edaa29421d6e56767b0ae6f1b835d10742", size = 76341929, upload-time = "2026-09-21T03:34:48.15Z" }, +] + [[package]] name = "nvidia-cutlass-dsl-libs-cu13" version = "4.5.1" @@ -5357,42 +6610,58 @@ wheels = [ [[package]] name = "nvidia-cutlass-dsl-libs-cu13" -version = "4.5.2" +version = "4.8.0" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32'", - "python_full_version >= '3.14' and sys_platform == 'emscripten'", - "python_full_version >= '3.14' and sys_platform == 'darwin'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", - "python_full_version == '3.13.*' and sys_platform == 'darwin'", - "python_full_version == '3.13.*' and sys_platform != 'darwin'", - "python_full_version == '3.12.*' and sys_platform == 'darwin'", - "python_full_version == '3.12.*' and sys_platform != 'darwin'", - "python_full_version == '3.11.*' and sys_platform == 'darwin'", - "python_full_version == '3.11.*' and sys_platform != 'darwin'", - "python_full_version < '3.11' and sys_platform == 'darwin'", - "python_full_version < '3.11' and sys_platform != 'darwin'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", ] dependencies = [ - { name = "cuda-python" }, - { name = "numpy" }, - { name = "nvidia-cutlass-dsl-libs-base", version = "4.5.2", source = { registry = "https://pypi.org/simple" } }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cuda-bindings", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cutlass-dsl-libs-base", version = "4.8.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x' and extra == 'extra-7-vec-inf-vllm') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/12/96/1519dc5fb936b2e8d519710a1134ecfd162dfbdb014f15ea4534f52ca221/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:696c65ca03995713b6719bc59b7df06f8ec1d263d7eb6ac77aa011201e142bd5", size = 79086862, upload-time = "2026-05-25T03:39:21.998Z" }, - { url = "https://files.pythonhosted.org/packages/3f/1e/12d1773571cd5f3cb2ff2a7570badfe9ccc1361e9f6684b17f7ff092c188/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:587494d0ab615b805fac86b43a3c1b855182f455681c9cc4ddb1b8973f44a7cc", size = 78759394, upload-time = "2026-05-25T03:42:32.448Z" }, - { url = "https://files.pythonhosted.org/packages/34/24/4ad875105f8b834ff0a6dce484c8ac124c292368338b087b993b70288385/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:f4a7b72147c2efdc7963c64475eac4ed67eb1dd5fdf5b0300daf79319fe9a38a", size = 79081923, upload-time = "2026-05-25T03:40:22.457Z" }, - { url = "https://files.pythonhosted.org/packages/3c/3d/2153608b1f8f594ccfc67daa45a1d0ff600b9e552b1e5662644e6e3ebec3/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:df61430d6110eea872acb39257042814bf02dcbb1f8d55ea0c5681bb7ce5836a", size = 78759970, upload-time = "2026-05-25T03:43:46.762Z" }, - { url = "https://files.pythonhosted.org/packages/21/e5/aeb570713a7bd6c2cb08102c2ebe6de234ef1bbc276d1af4643266cd71a8/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:3032405dff28892340f96b467e744a822079cae454dce534fc17b77e85190e42", size = 79084280, upload-time = "2026-05-25T03:40:57.547Z" }, - { url = "https://files.pythonhosted.org/packages/03/60/443e559139da15ab544761ac14f4206dffb981af48cc9856cd5b5b7cf0e7/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:80f0cd402e0f1d1571e5aed33bfa17dbc9cb90cc5b1352f0f806b4788558e80e", size = 78759198, upload-time = "2026-05-25T03:45:59.297Z" }, - { url = "https://files.pythonhosted.org/packages/98/57/bc7248c02c3e4ee2ed03e194ceda9861a46fa23f0da5140bd8060a086b1e/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:64e994554af4da59f75754b9df1a2b1bdfdb96b58c2457802da13d586fb58cde", size = 79086223, upload-time = "2026-05-25T03:41:27.363Z" }, - { url = "https://files.pythonhosted.org/packages/5c/9f/b7928ff505e577c1021c07b206ce32d285aae793763d524023c1800b6dc9/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:c7a5ce1c01616fc4c3ac492e011c543a79c3dde86aaf20a8af55e9d40ef2b2e6", size = 78759546, upload-time = "2026-05-25T03:45:25.834Z" }, - { url = "https://files.pythonhosted.org/packages/d1/65/a8e16a9647acef4f43ea2e046cd7eeb3e5779e89089c3939a5d25fe47f57/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:aabd41c980083db94950a4010c2c1ca156d4ab56701605739a3fba388ac9736b", size = 79087704, upload-time = "2026-05-25T03:41:57.202Z" }, - { url = "https://files.pythonhosted.org/packages/aa/83/d335575e1d37f6c436b1e3203ded6f352678937b9f30b900b643f9df0f9d/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:888edad4fe1e9b683fddcbc6969437527ccd0eb8740e60dce8f29f6a3a22c825", size = 78758331, upload-time = "2026-05-25T03:43:08.093Z" }, - { url = "https://files.pythonhosted.org/packages/89/7c/2bf50f2649f06a97a935919f71d2d0e40d7648364319b834548ed664d6d3/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:c4d3ea9080c5a92f8f4a69451ef7036f43bfc3d7f8a426dd70258f0e237c05fb", size = 79092929, upload-time = "2026-05-25T03:39:51.239Z" }, - { url = "https://files.pythonhosted.org/packages/36/80/8ced4c7e1ead8d1e3ac6c823db9e387dbcfd41232e11655d5bc94e950c75/nvidia_cutlass_dsl_libs_cu13-4.5.2-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:1d255f4a308eb0d228d2466a415a8489b8337db1d322f5d8428e60139b41a317", size = 78765148, upload-time = "2026-05-25T03:44:52.411Z" }, + { url = "https://files.pythonhosted.org/packages/ae/cd/83ab70c7ab233687c290de1ae9d756d2fadce6d770be6cc9b039e76ed786/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:02a5b208b79cc9dae294365c55b0cb1d5684bfdd220bd11eadeb230d0a75d05e", size = 91004241, upload-time = "2026-09-21T03:35:52.579Z" }, + { url = "https://files.pythonhosted.org/packages/ad/51/706b6af7efc60beaa2eb3dd8cd3f2bda6023cadd4dc6ab1948b2a1100707/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:55e036b14a4c999e097b498f73e8612e9ddc4017fdce0015fc1702f052c45c8e", size = 91952434, upload-time = "2026-09-21T03:36:16.694Z" }, + { url = "https://files.pythonhosted.org/packages/83/22/c7207e6ae2210460e37beccda585589b794428ee5007d2fa6734a82d21eb/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp310-cp310-win_amd64.whl", hash = "sha256:5170d10ea7f0cca7ef07f0c443476e913909d1840841c221f5fbaca3703d8b32", size = 75174142, upload-time = "2026-09-21T03:36:48.121Z" }, + { url = "https://files.pythonhosted.org/packages/3d/fd/39847d50b7965e6165657c1aadb9f505da3e3934c29ed34b2b1c07696ad1/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:ba1ec6d671f008152e73986893509676d0e0102ab242d533491814cde5c8e741", size = 90996933, upload-time = "2026-09-21T03:37:09.429Z" }, + { url = "https://files.pythonhosted.org/packages/37/51/839414c6711ffdf1ea626c9d0f12d29b417e51a3750f1d867aa333c980c5/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:1866559c8da5c3c0815047a1fb414e1745f47069b14bac5168885785b8553dc2", size = 91952020, upload-time = "2026-09-21T03:37:40.682Z" }, + { url = "https://files.pythonhosted.org/packages/a7/b0/c0a2e34f1484f9039a940199a6f3595b8b55f53117c9aa44ce2aed8660b5/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp311-cp311-win_amd64.whl", hash = "sha256:bc7c3a1bce134bde5cb99bf06c60cfc541a474e18c25fca2b85bea95d33deb4a", size = 75174580, upload-time = "2026-09-21T03:38:10.42Z" }, + { url = "https://files.pythonhosted.org/packages/bd/48/d5b62e1e3954dadc5e962d5be81c0c2acf3dcc7ed854f323447967ba3ec0/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:d61ebe2416cae8a2d5377889549df9cb5008af006a1692d7d006d8f747e62e6c", size = 90997691, upload-time = "2026-09-21T03:38:33.996Z" }, + { url = "https://files.pythonhosted.org/packages/df/75/6c52c58fc0fdab223feb9162886736390a6f902183d24df1508ac620a69d/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:463cd7f0c452ffeeecb530a15aeb1b10cbbd2c8da6c5aa2fecf65f836258a732", size = 91950801, upload-time = "2026-09-21T03:38:58.38Z" }, + { url = "https://files.pythonhosted.org/packages/fa/45/ddf311407c257ea57fb40a2602261d26f49073ef4c8e7f48e2a61934085c/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp312-cp312-win_amd64.whl", hash = "sha256:8684f4fef26bedab100931c8f65098e2aa3c3b0e9d730fad3b92780ad2773634", size = 75186004, upload-time = "2026-09-21T03:39:28.825Z" }, + { url = "https://files.pythonhosted.org/packages/40/9b/67b119b3eebc97d8fbf5414ef5f6b3dd31712719937dc1460feea2b084d4/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:5741a783dc09e294dada6844c9b8df8ff14fcc19db20be482055911a4fb9b73f", size = 90995158, upload-time = "2026-09-21T03:39:52.478Z" }, + { url = "https://files.pythonhosted.org/packages/9e/49/333deaacb1e7cd7f897d15a575eaab45e98338ea2346ca30edcac938da7c/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:41378cfa94ef3e74084d5157064ea807a92096123d378697a2d58bdbd582240a", size = 91952018, upload-time = "2026-09-21T03:40:11.246Z" }, + { url = "https://files.pythonhosted.org/packages/23/a2/fbb336dd4df0ca794282b792016c43710fad5804245cf20ee8e991a8d557/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp313-cp313-win_amd64.whl", hash = "sha256:cc92f1c5e9bdaa5c6d7e5d0e3a859b3364ae3c5b82726e1d1951d2ee1ad73d14", size = 75183701, upload-time = "2026-09-21T03:40:48.132Z" }, + { url = "https://files.pythonhosted.org/packages/16/cf/295081c6fa3289dcb66f8f78e8e2ce26e0e41fb71c4b5adda37d78267e74/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:bbdb2b14a74a35d382a30bee6f17e31faf0aabb6d87909b2b99c7d134a6cac10", size = 90998930, upload-time = "2026-09-21T03:41:18.056Z" }, + { url = "https://files.pythonhosted.org/packages/6e/ea/d1205eb6f6aa5b2c1437f0953873c81907f1a18d9409b38226c7819b0c15/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:09d17524a55f40df411616a771b7537be8ee056612681abc705424eebb30ba8e", size = 91951440, upload-time = "2026-09-21T03:41:40.56Z" }, + { url = "https://files.pythonhosted.org/packages/a7/2b/29bcd27920d6d8a1a9d835e35acdcc3e4288740286522ea5cf5a848932bc/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp314-cp314-win_amd64.whl", hash = "sha256:95bd45c1e50b387b2d065b45d87b1ed40a61997ba48f0c1c498b5292a2a5f845", size = 77186418, upload-time = "2026-09-21T03:42:00.501Z" }, + { url = "https://files.pythonhosted.org/packages/fe/db/d552e5b7866a269cd8a1dab73cd8e03df946ee0c801b35195b607aebad76/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:e73a78a79304e71d3832bdfbcbb52a5245e28fe49a8a1a61dbebbfca562184c0", size = 91011670, upload-time = "2026-09-21T03:42:24.743Z" }, + { url = "https://files.pythonhosted.org/packages/76/9a/8c0cf995e2949ccfcffbab7f667407011eef384b5812e56d0e3541992763/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:4bd241512a43bb936e34a76fafe9b13f8c282f0fa6041e2cc5fcb6cab47f5097", size = 91961820, upload-time = "2026-09-21T03:42:44.376Z" }, + { url = "https://files.pythonhosted.org/packages/f0/3b/847b6f7d0aa24421b75582fb53bf17ea837de2139e74db552686c06300aa/nvidia_cutlass_dsl_libs_cu13-4.8.0-cp314-cp314t-win_amd64.whl", hash = "sha256:74da9cd5ed4c9407d1963a0456107304f26db76b890b2bbd2d057fa5822101d6", size = 77302182, upload-time = "2026-09-21T03:43:12.675Z" }, ] [[package]] @@ -5408,11 +6677,41 @@ wheels = [ name = "nvidia-nccl-cu13" version = "2.28.9" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] wheels = [ { url = "https://files.pythonhosted.org/packages/39/55/1920646a2e43ffd4fc958536b276197ed740e9e0c54105b4bb3521591fc7/nvidia_nccl_cu13-2.28.9-py3-none-manylinux_2_18_aarch64.whl", hash = "sha256:01c873ba1626b54caa12272ed228dc5b2781545e0ae8ba3f432a8ef1c6d78643", size = 196561677, upload-time = "2025-11-18T05:49:03.45Z" }, { url = "https://files.pythonhosted.org/packages/b0/b4/878fefaad5b2bcc6fcf8d474a25e3e3774bc5133e4b58adff4d0bca238bc/nvidia_nccl_cu13-2.28.9-py3-none-manylinux_2_18_x86_64.whl", hash = "sha256:e4553a30f34195f3fa1da02a6da3d6337d28f2003943aa0a3d247bbc25fefc42", size = 196493177, upload-time = "2025-11-18T05:49:17.677Z" }, ] +[[package]] +name = "nvidia-nccl-cu13" +version = "2.29.7" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/72/0d/daf50d44177ee0cbc7ff0a0c91eb5ff676c82be42f9a970bc7597f440c3a/nvidia_nccl_cu13-2.29.7-py3-none-manylinux_2_18_aarch64.whl", hash = "sha256:674a12383e3c38a1bcccae7d4f3633b37852230b6047883cb2f4c2d1b36d9bf5", size = 206014712, upload-time = "2026-03-03T05:34:20.843Z" }, + { url = "https://files.pythonhosted.org/packages/67/f4/58e4e91b6919367c7aafb8e36fce9aad1a3047e536bf7e2fd560927d3a4c/nvidia_nccl_cu13-2.29.7-py3-none-manylinux_2_18_x86_64.whl", hash = "sha256:edd81538446786ec3b73972543e53bb43bcaf0bfc8ef76cb679fcc390ffe136d", size = 205976000, upload-time = "2026-03-03T05:36:24.472Z" }, +] + [[package]] name = "nvidia-nvjitlink" version = "13.0.88" @@ -5420,6 +6719,7 @@ source = { registry = "https://pypi.org/simple" } wheels = [ { url = "https://files.pythonhosted.org/packages/56/7a/123e033aaff487c77107195fa5a2b8686795ca537935a24efae476c41f05/nvidia_nvjitlink-13.0.88-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl", hash = "sha256:13a74f429e23b921c1109976abefacc69835f2f433ebd323d3946e11d804e47b", size = 40713933, upload-time = "2025-09-04T08:35:43.553Z" }, { url = "https://files.pythonhosted.org/packages/ab/2c/93c5250e64df4f894f1cbb397c6fd71f79813f9fd79d7cd61de3f97b3c2d/nvidia_nvjitlink-13.0.88-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:e931536ccc7d467a98ba1d8b89ff7fa7f1fa3b13f2b0069118cd7f47bff07d0c", size = 38768748, upload-time = "2025-09-04T08:35:20.008Z" }, + { url = "https://files.pythonhosted.org/packages/e4/01/07530b0e37546231052e30234540289c42eaffa486f1a34a87fed340157b/nvidia_nvjitlink-13.0.88-py3-none-win_amd64.whl", hash = "sha256:634e96e3da9ef845ae744097a1f289238ecf946ce0b82e93cdce14b9782e682f", size = 36035115, upload-time = "2025-09-04T08:43:03.001Z" }, ] [[package]] @@ -5438,6 +6738,7 @@ source = { registry = "https://pypi.org/simple" } wheels = [ { url = "https://files.pythonhosted.org/packages/c2/f3/d86c845465a2723ad7e1e5c36dcd75ddb82898b3f53be47ebd429fb2fa5d/nvidia_nvtx-13.0.85-py3-none-manylinux1_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:4936d1d6780fbe68db454f5e72a42ff64d1fd6397df9f363ae786930fd5c1cd4", size = 148047, upload-time = "2025-09-04T08:29:01.761Z" }, { url = "https://files.pythonhosted.org/packages/a8/64/3708a90d1ebe202ffdeb7185f878a3c84d15c2b2c31858da2ce0583e2def/nvidia_nvtx-13.0.85-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:cb7780edb6b14107373c835bf8b72e7a178bac7367e23da7acb108f973f157a6", size = 148878, upload-time = "2025-09-04T08:28:53.627Z" }, + { url = "https://files.pythonhosted.org/packages/d2/50/0e2220f8620a177de994211186ffc5bfa9f2ce1e1282797f8f90096f9f88/nvidia_nvtx-13.0.85-py3-none-win_amd64.whl", hash = "sha256:d66ea44254dd3c6eacc300047af6e1288d2269dd072b417e0adffbf479e18519", size = 137066, upload-time = "2025-09-04T08:39:25.649Z" }, ] [[package]] @@ -5450,6 +6751,35 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/8f/96/4de7a37803d168337ab36f81ecbc496c7c21c9b06ec68ce0ecc381af88d4/nvidia_nvvm-13.3.33-py3-none-win_amd64.whl", hash = "sha256:b1c63cf8972d8a1ff153c5ac4cc7038fe6ef705aa38415f12007b0e5e4c31b79", size = 60175824, upload-time = "2026-05-26T17:13:27.588Z" }, ] +[[package]] +name = "nvtx" +version = "0.2.15" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/92/dd/692765e87de30bae1522cdffaa0f2b52949658a92a0fa6d96b1a01eae9d2/nvtx-0.2.15.tar.gz", hash = "sha256:2287d3be05b85661deb386f878d1f536c2e532774aa9ec7a50c434942ed81ae5", size = 121230, upload-time = "2026-03-18T10:01:25.547Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/9d/ef/ea1e9d92afd07fdf2a2390e508f1d214e5ba890561d7849d6ca708534b9d/nvtx-0.2.15-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:a4f50832fd90a1b480a9deef6e4cd48015b61869095b54dd1a7afe87b4138c6a", size = 768543, upload-time = "2026-03-18T10:07:21.819Z" }, + { url = "https://files.pythonhosted.org/packages/32/8e/b42c05cf3cc43c51f21fdda6f7c4fe28a595c6d2bdb0cfbf0477dc5805f2/nvtx-0.2.15-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5f3362f0db4252514719326c9d5662b0f93d254659ba97b9c8dbe556286e0e3e", size = 771975, upload-time = "2026-03-18T10:12:23.772Z" }, + { url = "https://files.pythonhosted.org/packages/60/77/fc000055b5bb1651cdd772f0fe1fd9a16c7773b28dfc5624eea331d1415d/nvtx-0.2.15-cp310-cp310-win_amd64.whl", hash = "sha256:d71f934e580d4572f382712b6da464ab69e4c212981506f781f927d5c6d935d6", size = 134503, upload-time = "2026-03-18T10:04:05.773Z" }, + { url = "https://files.pythonhosted.org/packages/80/65/435d10b2041ee082c07d5aed129afd504012c8908796d695f10e66bcc716/nvtx-0.2.15-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:157b80ea9b4db6c8f47f8dbe2fa2e81e7a7f1445bb87f8268f43dec9210b78a1", size = 806443, upload-time = "2026-03-18T10:05:49.308Z" }, + { url = "https://files.pythonhosted.org/packages/47/bc/be94576ba33af75bcc68a857daade64cb86481764d4fb0f36308b1f6fc85/nvtx-0.2.15-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:02bca69ee55e0be41eabf908de9dbcdd18e702c7f49f9aa63fd396ce684ff5d5", size = 808183, upload-time = "2026-03-18T10:11:16.262Z" }, + { url = "https://files.pythonhosted.org/packages/f6/7a/42109f1cfb1ff9913201cb2b804956a4f003db4c018c2522a3c8066b3a1c/nvtx-0.2.15-cp311-cp311-win_amd64.whl", hash = "sha256:dbe41f78f5a811bd4cdad0a237e5b41a4937d8c2c6c9abdd161091671a598bc0", size = 134631, upload-time = "2026-03-18T10:02:11.247Z" }, + { url = "https://files.pythonhosted.org/packages/c2/07/698355285a03a366ef63ea9762fc1feef3f9f25483e1655408f72d827090/nvtx-0.2.15-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:2cc530cd0f1a2c14a3a7e683833db509888ac5ed4ead94e5c9e2c7317c6937a7", size = 807159, upload-time = "2026-03-18T10:09:49.232Z" }, + { url = "https://files.pythonhosted.org/packages/c0/d1/08f22448d83481408d663065764ba583df091a7de629ed38fc97e522f1af/nvtx-0.2.15-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:3ca8030a6d197952318013dd1c12c22da1d4b9feb76ba72e0fcd449961183c2c", size = 806187, upload-time = "2026-03-18T10:13:32.972Z" }, + { url = "https://files.pythonhosted.org/packages/54/23/c97c39e3b7ba256aa343cb828ca0d1c8421f705ca84795658ecd14ca95ed/nvtx-0.2.15-cp312-cp312-win_amd64.whl", hash = "sha256:70a1e768964e0520b68ccabc4df391cc227537c45936a7eba6507bc65e617e00", size = 129178, upload-time = "2026-03-18T10:02:55.299Z" }, + { url = "https://files.pythonhosted.org/packages/05/c9/8341224b8284f7deb6a634119939de5885adc421e64b6743693b30da2186/nvtx-0.2.15-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d28660d9c46f8ba750d781572b6aa5a1e6221abba224ab32d7fb32c2d0fd67df", size = 780787, upload-time = "2026-03-18T10:10:40.634Z" }, + { url = "https://files.pythonhosted.org/packages/b1/c0/4a5bb7897918de7c7e0191d9342df8ae4cb797ff07276e0f20d13e497ce7/nvtx-0.2.15-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:10749686633f880ad53dcdbb2179fad41b45dcf5b7631d4a1070a577577bd386", size = 782575, upload-time = "2026-03-18T10:13:57.3Z" }, + { url = "https://files.pythonhosted.org/packages/38/b9/6b381ac7c5a3ded331aebbf25f8959d19b51d320fb2514c76c6b6edddaaa/nvtx-0.2.15-cp313-cp313-win_amd64.whl", hash = "sha256:a6650b029263d12f8427a4dee8bd59cb9c91bccb60543bfcb20bc2b00fdcd672", size = 128764, upload-time = "2026-03-18T10:02:33.343Z" }, + { url = "https://files.pythonhosted.org/packages/75/69/a9acb6d95d2e0e381b2956544768528dd8d7a9e827af8c2014169d838284/nvtx-0.2.15-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:25813ead4fff4d3a6e04f69a72507b096a6bdbecefa369f1100b0e584767bca8", size = 833375, upload-time = "2026-03-18T10:06:31.955Z" }, + { url = "https://files.pythonhosted.org/packages/38/56/c7e8645061cc2fc23f3a54f33e1e340df59216f07dcfb97d46b8ae7dd26c/nvtx-0.2.15-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:3741edac4678b92f03d22a3f0a2dfd469f422f85e63db71b038e02525b2404ad", size = 788639, upload-time = "2026-03-18T10:12:01.69Z" }, + { url = "https://files.pythonhosted.org/packages/96/03/fadd82acdbca6d1c49ac517081a0c3714346f52f4c7e1d4449d77605b4aa/nvtx-0.2.15-cp313-cp313t-win_amd64.whl", hash = "sha256:8be06c3c8c267eba56a0396366b9593092e0b75ea8d3702b303d48c0a1662f0e", size = 142609, upload-time = "2026-03-18T10:01:48.832Z" }, + { url = "https://files.pythonhosted.org/packages/e0/5b/ca0ba6fa769d08174b7a5b4775c279e2e26611cdd5e7833aa699187871c7/nvtx-0.2.15-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b5171b8283dd3ea9ae688a86d16901b4c2c142c4eb0a4bdbf6c222f5f67f9524", size = 781769, upload-time = "2026-03-18T10:08:59.357Z" }, + { url = "https://files.pythonhosted.org/packages/f7/e1/e02fafc01c18f1868a2d2c030953f49e38d65f2d95884789a6c46ff308f1/nvtx-0.2.15-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:3c6d0f27d4f8a2f479eb64a6b842c13aee32120348a1715d995b9bb9f75b35cf", size = 774614, upload-time = "2026-03-18T10:12:46.979Z" }, + { url = "https://files.pythonhosted.org/packages/20/77/a2b64335bab7c75fe1c054cc4ebe2d3b3234cbdb04d2e1d6ca73551c54f5/nvtx-0.2.15-cp314-cp314-win_amd64.whl", hash = "sha256:9934fad0b441cfa6e896a848b092498ba23e2ff205c2b9a7b60520ff8367ffef", size = 130932, upload-time = "2026-03-18T10:03:43.507Z" }, + { url = "https://files.pythonhosted.org/packages/db/24/528619230976c18364eda2340906ea67b3bf7588b7ce59e054723614abae/nvtx-0.2.15-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:aca61135c76b8107ae3c994325613afa661e1336a991c59cc9c6176829b3b32c", size = 834439, upload-time = "2026-03-18T10:05:01.181Z" }, + { url = "https://files.pythonhosted.org/packages/ef/7b/c1b96f13ef89bdf2a8c2f326a97bed89699271990d7c8624fda3fedc6e61/nvtx-0.2.15-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:58653bf6fd8453947b9e5153da2ad7aeb0ceafa030de7f133efb3eada5da7ca7", size = 790247, upload-time = "2026-03-18T10:11:39.124Z" }, + { url = "https://files.pythonhosted.org/packages/14/5d/e000de781d92b732d52c572517db0e9e3a0085795f8bdc18201713c52d1f/nvtx-0.2.15-cp314-cp314t-win_amd64.whl", hash = "sha256:9d1d10db4fb4a3b0ffd6ed37bf25f0a966a3b4d34b3c9abb1f6572732959a6e5", size = 149109, upload-time = "2026-03-18T10:03:21.615Z" }, +] + [[package]] name = "openai" version = "2.6.1" @@ -5488,17 +6818,27 @@ name = "openai" version = "2.31.0" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", @@ -5525,18 +6865,6 @@ wheels = [ name = "openai-harmony" version = "0.0.4" source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'darwin'", - "python_full_version >= '3.14' and sys_platform != 'darwin'", - "python_full_version == '3.13.*' and sys_platform == 'darwin'", - "python_full_version == '3.13.*' and sys_platform != 'darwin'", - "python_full_version == '3.12.*' and sys_platform == 'darwin'", - "python_full_version == '3.12.*' and sys_platform != 'darwin'", - "python_full_version == '3.11.*' and sys_platform == 'darwin'", - "python_full_version == '3.11.*' and sys_platform != 'darwin'", - "python_full_version < '3.11' and sys_platform == 'darwin'", - "python_full_version < '3.11' and sys_platform != 'darwin'", -] dependencies = [ { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, @@ -5559,46 +6887,8 @@ wheels = [ ] [[package]] -name = "openai-harmony" -version = "0.0.8" -source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32'", - "python_full_version >= '3.14' and sys_platform == 'emscripten'", - "python_full_version >= '3.14' and sys_platform == 'darwin'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", - "python_full_version == '3.13.*' and sys_platform == 'darwin'", - "python_full_version == '3.13.*' and sys_platform != 'darwin'", - "python_full_version == '3.12.*' and sys_platform == 'darwin'", - "python_full_version == '3.12.*' and sys_platform != 'darwin'", - "python_full_version == '3.11.*' and sys_platform == 'darwin'", - "python_full_version == '3.11.*' and sys_platform != 'darwin'", - "python_full_version < '3.11' and sys_platform == 'darwin'", - "python_full_version < '3.11' and sys_platform != 'darwin'", -] -dependencies = [ - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/3e/92/2d038d096f29179c7c9571b431f9e739f87a487121901725e23fe338dd9d/openai_harmony-0.0.8.tar.gz", hash = "sha256:6e43f98e6c242fa2de6f8ea12eab24af63fa2ed3e89c06341fb9d92632c5cbdf", size = 284777, upload-time = "2025-11-05T19:07:06.727Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/45/c6/2502f416d46be3ec08bb66d696cccffb57781a499e3ff2e4d7c174af4e8f/openai_harmony-0.0.8-cp38-abi3-macosx_11_0_arm64.whl", hash = "sha256:029ec25ca74abe48fdb58eb9fdd2a8c1618581fc33ce8e5653f8a1ffbfbd9326", size = 2627806, upload-time = "2025-11-05T19:06:57.063Z" }, - { url = "https://files.pythonhosted.org/packages/d3/d2/ce6953ca87db9cae3e775024184da7d1c5cb88cead19a2d75b42f00a959c/openai_harmony-0.0.8-cp38-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e4f709815924ec325b9a890e6ab2bbb0ceec8e319a4e257328eb752cf36b2efc", size = 2948463, upload-time = "2025-11-05T19:06:48.17Z" }, - { url = "https://files.pythonhosted.org/packages/fa/4c/b553c9651662d6ce102ca7f3629d268b23df1abe5841e24bed81e8a8e949/openai_harmony-0.0.8-cp38-abi3-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:5cfcfd963b50a41fc656c84d3440ca6eecdccd6c552158ce790b8f2e33dfb5a9", size = 2704083, upload-time = "2025-11-05T19:06:50.205Z" }, - { url = "https://files.pythonhosted.org/packages/9b/af/4eec8f9ab9c27bcdb444460c72cf43011d176fc44c79d6e113094ca1e152/openai_harmony-0.0.8-cp38-abi3-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:0a3a16972aa1cee38ea958470cd04ac9a2d5ac38fdcf77ab686611246220c158", size = 2959765, upload-time = "2025-11-05T19:06:53.62Z" }, - { url = "https://files.pythonhosted.org/packages/11/3c/33f3374e4624e0e776f6b13b73c45a7ead7f9c4529f8369ed5bfcaa30cac/openai_harmony-0.0.8-cp38-abi3-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:b4d5cfa168e74d08f8ba6d58a7e49bc7daef4d58951ec69b66b0d56f4927a68d", size = 3427031, upload-time = "2025-11-05T19:06:51.829Z" }, - { url = "https://files.pythonhosted.org/packages/25/3f/1a192b93bb47c6b44cd98ba8cc1d3d2a9308f1bb700c3017e6352da11bda/openai_harmony-0.0.8-cp38-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c007d277218a50db8839e599ed78e0fffe5130f614c3f6d93ae257f282071a29", size = 2953260, upload-time = "2025-11-05T19:06:55.406Z" }, - { url = "https://files.pythonhosted.org/packages/5b/f8/93b582cad3531797c3db7c2db5400fd841538ccddfd9f5e3df61be99a630/openai_harmony-0.0.8-cp38-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:8565d4f5a0638da1bffde29832ed63c9e695c558611053add3b2dc0b56c92dbc", size = 3127044, upload-time = "2025-11-05T19:06:59.553Z" }, - { url = "https://files.pythonhosted.org/packages/1d/10/4327dbf87f75ae813405fd9a9b4a5cde63d506ffed0a096a440a4cabd89c/openai_harmony-0.0.8-cp38-abi3-musllinux_1_2_armv7l.whl", hash = "sha256:cbaa3bda75ef0d8836e1f8cc84af62f971b1d756d740efc95c38c3e04c0bfde2", size = 2932931, upload-time = "2025-11-05T19:07:01.437Z" }, - { url = "https://files.pythonhosted.org/packages/8a/c8/1774eec4f6f360ef57618fb8f52e3d3af245b2491bd0297513aa09eec04b/openai_harmony-0.0.8-cp38-abi3-musllinux_1_2_i686.whl", hash = "sha256:772922a9bd24e133950fad71eb1550836f415a88e8c77870e12d0c3bd688ddc2", size = 2996140, upload-time = "2025-11-05T19:07:03.438Z" }, - { url = "https://files.pythonhosted.org/packages/60/c3/3d1e01e2dba517a91760e4a03e4f20ffc75039a6fe584d0e6f9b5c78fd15/openai_harmony-0.0.8-cp38-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:007b0476a1f331f8130783f901f1da6f5a7057af1a4891f1b6a31dec364189b5", size = 3205080, upload-time = "2025-11-05T19:07:05.078Z" }, - { url = "https://files.pythonhosted.org/packages/14/63/119de431572d7c70a7bf1037034a9be6ed0a7502a7498ba7302bca5b3242/openai_harmony-0.0.8-cp38-abi3-win32.whl", hash = "sha256:a9b5f893326b28d9e935ade14b4f655f5a840942473bc89b201c25f7a15af9cf", size = 2082457, upload-time = "2025-11-05T19:07:09.631Z" }, - { url = "https://files.pythonhosted.org/packages/40/1f/c83cf5a206c263ee70448a5ae4264682555f4d0b5bed0d2cc6ca1108103d/openai_harmony-0.0.8-cp38-abi3-win_amd64.whl", hash = "sha256:39d44f0d8f466bd56698e7ead708bead3141e27b9b87e3ab7d5a6d0e4a869ee5", size = 2438369, upload-time = "2025-11-05T19:07:08.1Z" }, -] - -[[package]] -name = "opencensus" -version = "0.11.4" +name = "opencensus" +version = "0.11.4" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "google-api-core" }, @@ -5624,7 +6914,7 @@ name = "opencv-python-headless" version = "4.13.0.92" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy" }, + { name = "numpy", marker = "python_full_version >= '3.11'" }, ] wheels = [ { url = "https://files.pythonhosted.org/packages/79/42/2310883be3b8826ac58c3f2787b9358a2d46923d61f88fedf930bc59c60c/opencv_python_headless-4.13.0.92-cp37-abi3-macosx_13_0_arm64.whl", hash = "sha256:1a7d040ac656c11b8c38677cc8cccdc149f98535089dbe5b081e80a4e5903209", size = 46247192, upload-time = "2026-02-05T07:01:35.187Z" }, @@ -5642,7 +6932,7 @@ name = "opentelemetry-api" version = "1.45.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/2e/02/6e0ae9cc61bd3169d401077b507b3ebc344745171e1051ab430be012dcd9/opentelemetry_api-1.45.1.tar.gz", hash = "sha256:aa38ed19bcc084ba42782a73255b3582283eced7ad6dddbd6695189e69adfb75", size = 72804, upload-time = "2026-10-06T17:32:58.133Z" } @@ -5672,8 +6962,8 @@ name = "opentelemetry-exporter-otlp" version = "1.45.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "opentelemetry-exporter-otlp-proto-grpc" }, - { name = "opentelemetry-exporter-otlp-proto-http" }, + { name = "opentelemetry-exporter-otlp-proto-grpc", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-exporter-otlp-proto-http", marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/e3/6f/5a561048ea372894f22f58f70e8478e4cd97d9a4e2d6f014559584a29bc5/opentelemetry_exporter_otlp-1.45.1.tar.gz", hash = "sha256:d0ac35592e77663a9fabf2740b4818c57196b1b764e0c8449d0d7bb2c7b2bc67", size = 6094, upload-time = "2026-10-06T17:33:00.966Z" } wheels = [ @@ -5709,15 +6999,14 @@ name = "opentelemetry-exporter-otlp-proto-grpc" version = "1.45.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "googleapis-common-protos" }, - { name = "grpcio" }, - { name = "opentelemetry-api" }, - { name = "opentelemetry-exporter-otlp-common" }, - { name = "opentelemetry-exporter-otlp-proto-common" }, - { name = "opentelemetry-proto" }, - { name = "opentelemetry-sdk" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "googleapis-common-protos", marker = "python_full_version >= '3.11'" }, + { name = "grpcio", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-api", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-exporter-otlp-common", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-exporter-otlp-proto-common", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-proto", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-sdk", marker = "python_full_version >= '3.11'" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/d6/00/a82af0be959dc58495740b169c6669a86e0811f6cd353a01eda34d255db3/opentelemetry_exporter_otlp_proto_grpc-1.45.1.tar.gz", hash = "sha256:3b3dcfbfdcb4e35149fcf309972282054b45228f5c10547d0095d6578510a9a0", size = 25997, upload-time = "2026-10-06T17:33:05.114Z" } wheels = [ @@ -5731,14 +7020,13 @@ source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "googleapis-common-protos" }, { name = "opentelemetry-api" }, - { name = "opentelemetry-exporter-http-transport", extra = ["requests"], marker = "extra == 'extra-7-vec-inf-vllm'" }, + { name = "opentelemetry-exporter-http-transport", extra = ["requests"] }, { name = "opentelemetry-exporter-otlp-common" }, { name = "opentelemetry-exporter-otlp-proto-common" }, { name = "opentelemetry-proto" }, { name = "opentelemetry-sdk" }, { name = "requests" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" } }, ] sdist = { url = "https://files.pythonhosted.org/packages/1b/17/26487707ea4caa97b17e6e4b5fa72133a53512ffa2f5cf7a49ef284b29cb/opentelemetry_exporter_otlp_proto_http-1.45.1.tar.gz", hash = "sha256:45c218405ce3fd879596924b1874bf9a8f6880206d61065c5a912c8e5c297fb7", size = 28839, upload-time = "2026-10-06T17:33:05.713Z" } wheels = [ @@ -5764,7 +7052,8 @@ name = "opentelemetry-proto" version = "1.45.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "protobuf" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/4b/7f/15f014fb195da6c2dbb6c71399b8e76824878718e94de6454038488eed28/opentelemetry_proto-1.45.1.tar.gz", hash = "sha256:79e0fb95e4616691a469439238aa9224d75779b3e108e895d1aa125ab29ca77c", size = 46488, upload-time = "2026-10-06T17:33:11.49Z" } wheels = [ @@ -5776,9 +7065,9 @@ name = "opentelemetry-sdk" version = "1.45.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "opentelemetry-api" }, - { name = "opentelemetry-semantic-conventions" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "opentelemetry-api", marker = "python_full_version >= '3.11' or extra == 'extra-7-vec-inf-vllm'" }, + { name = "opentelemetry-semantic-conventions", marker = "python_full_version >= '3.11' or extra == 'extra-7-vec-inf-vllm'" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/a1/79/7392e21a1c8f0c61d90b223e31c7e48cb9d452e91a6b820ad24cca5f23c4/opentelemetry_sdk-1.45.1.tar.gz", hash = "sha256:63d24a6ca645019a631e6a51999c73e93adcac1196ca640b8ae78a7cc4762bf3", size = 218324, upload-time = "2026-10-06T17:33:13.26Z" } @@ -5791,8 +7080,8 @@ name = "opentelemetry-semantic-conventions" version = "0.66b1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "opentelemetry-api" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "opentelemetry-api", marker = "python_full_version >= '3.11' or extra == 'extra-7-vec-inf-vllm'" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/46/e4/dbbfb2a010c4db2224a5114638acede6fe563d33cc20fb1752cebcbe6298/opentelemetry_semantic_conventions-0.66b1.tar.gz", hash = "sha256:497ca63bf383723411e8eaf60c8779e9877633c936bb641080adab59d0eb6ec8", size = 150250, upload-time = "2026-10-06T17:33:14.073Z" } @@ -5805,8 +7094,8 @@ name = "opentelemetry-semantic-conventions-ai" version = "0.5.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "opentelemetry-sdk" }, - { name = "opentelemetry-semantic-conventions" }, + { name = "opentelemetry-sdk", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-semantic-conventions", marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/24/02/10aeacc37a38a3a8fa16ff67bec1ae3bf882539f6f9efb0f70acf802ca2d/opentelemetry_semantic_conventions_ai-0.5.1.tar.gz", hash = "sha256:153906200d8c1d2f8e09bd78dbef526916023de85ac3dab35912bfafb69ff04c", size = 26533, upload-time = "2026-03-26T14:20:38.73Z" } wheels = [ @@ -5903,6 +7192,30 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/c0/d1/facb5b5051fabb0ef9d26c6544d87ef19a939a9a001198655d0d891062dd/orjson-3.11.8-cp314-cp314-win_arm64.whl", hash = "sha256:6ccdea2c213cf9f3d9490cbd5d427693c870753df41e6cb375bd79bcbafc8817", size = 127330, upload-time = "2026-03-31T16:16:25.496Z" }, ] +[[package]] +name = "oss-harmony" +version = "0.0.11" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/43/88/ca56e5b17df4763bbb552c7be3abe432b51db3c51f6327a3848b88ad7c08/oss_harmony-0.0.11.tar.gz", hash = "sha256:ccad467b71d61490df1f2204b7a064990833338615ac135536c5b3a8e2d03a15", size = 2198780, upload-time = "2026-05-13T20:49:11.767Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/8b/4c/400a7e375943ff06447c8cd3920819d2e37c808d0cfea3296b159e961610/oss_harmony-0.0.11-cp38-abi3-macosx_11_0_arm64.whl", hash = "sha256:83ca71b85be6e2f8e93527673cd2eff4da9180608ee5d4197bd4f01afc4fa6be", size = 3337401, upload-time = "2026-05-13T20:49:00.051Z" }, + { url = "https://files.pythonhosted.org/packages/fd/e9/0b33d4c592e1422c9faeaf1cfe10731328ecf3504e900dc7921109affd55/oss_harmony-0.0.11-cp38-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f1f74ef7f8be32fa1dcfe1644839af5f0873acce78b5668c2aa39f26d83218ff", size = 3493239, upload-time = "2026-05-13T20:48:49.588Z" }, + { url = "https://files.pythonhosted.org/packages/de/c3/502b7ba6f36645117b58859b250355cba34d7107ec104ea40087b752977b/oss_harmony-0.0.11-cp38-abi3-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:2221f2790411cfd5d0e517769dc2662b15b2711e44728291961506a367ffc3d1", size = 3415967, upload-time = "2026-05-13T20:48:51.731Z" }, + { url = "https://files.pythonhosted.org/packages/8b/f7/f802d7aba027cd4e1c0ff5509c645de8d70fa674b31d80cdf8fec63ac844/oss_harmony-0.0.11-cp38-abi3-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f215934b05f3a7658e0cf30cf92bc25f4aed125a693529c53dd97325ecfbf7ef", size = 3707786, upload-time = "2026-05-13T20:48:53.91Z" }, + { url = "https://files.pythonhosted.org/packages/89/68/001d7429a692b2db0c03bd49b756f834716b64afc0bbbece96824e8f0af8/oss_harmony-0.0.11-cp38-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d877c6b526cdbdf55068bae8f05c60398e929c9e296c7f69a1d533339708be80", size = 3567858, upload-time = "2026-05-13T20:48:57.904Z" }, + { url = "https://files.pythonhosted.org/packages/3b/5d/edf954b0aace872ec137a9ef7b92c29120bd16a4277fd11c5b83e7d1ed16/oss_harmony-0.0.11-cp38-abi3-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:e451e5dc28b0d883f33aa17a77c923afcab5d26ef3b95d7593b68a07e399706e", size = 3633793, upload-time = "2026-05-13T20:48:56.037Z" }, + { url = "https://files.pythonhosted.org/packages/4b/c9/720aed1527d87f82e07a7f35c3c81c6d9198a8d1223bdbd298b3c66a72db/oss_harmony-0.0.11-cp38-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:036505bfb39c6b50edcf717349518d269bb375207f7f8e95f517e938b8f2e0c8", size = 3695703, upload-time = "2026-05-13T20:49:02.857Z" }, + { url = "https://files.pythonhosted.org/packages/94/a8/fb8cad33079ee848514ea70688005ef7b6c32ccee788cc18fcb04e24fa74/oss_harmony-0.0.11-cp38-abi3-musllinux_1_2_armv7l.whl", hash = "sha256:952218411c219dc7ca8ee56eb86271d015d0c347896b2009906096422d7a94cf", size = 3703138, upload-time = "2026-05-13T20:49:04.937Z" }, + { url = "https://files.pythonhosted.org/packages/0d/7d/d71b3ee5abfbf9130eea646b9589e8f00a039514cdca2ab6dbffa22f8351/oss_harmony-0.0.11-cp38-abi3-musllinux_1_2_i686.whl", hash = "sha256:5c7b0cea27e1e4e9fc485bf8039a119d6732718cce6b067d1f7de9e637572348", size = 3790649, upload-time = "2026-05-13T20:49:07.337Z" }, + { url = "https://files.pythonhosted.org/packages/48/9a/ad24d20ea1e8b3167744f5d357d48a116086258ca58540986a0896781864/oss_harmony-0.0.11-cp38-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:0d1545203973abb3002dcc6c5cffd14ca4a376b655c14ce190c78eb6aaf5f6fd", size = 3807882, upload-time = "2026-05-13T20:49:09.767Z" }, + { url = "https://files.pythonhosted.org/packages/03/33/2917125287b46255cfc6d7e0bc4bc5e5c90f7ad4e02d3922c2f0f95b696b/oss_harmony-0.0.11-cp38-abi3-win32.whl", hash = "sha256:fd91609eef825f342ae7ac990651cfebcf28aac6b6ad1186bf37c5b593cdf950", size = 3135714, upload-time = "2026-05-13T20:49:15.924Z" }, + { url = "https://files.pythonhosted.org/packages/c8/c9/eedd8d02be8bbaa586f8a323de2fc976e6053582411020b7be1f7c7e1734/oss_harmony-0.0.11-cp38-abi3-win_amd64.whl", hash = "sha256:4db175b6f6264cd60ceb692c0e686937e072ca6ea1239d0d4cace2cfb50f4974", size = 3266750, upload-time = "2026-05-13T20:49:13.404Z" }, +] + [[package]] name = "outlines" version = "0.1.11" @@ -5923,7 +7236,7 @@ dependencies = [ { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "referencing" }, { name = "requests" }, - { name = "torch" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" } }, { name = "tqdm" }, { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, @@ -5980,18 +7293,26 @@ name = "outlines-core" version = "0.2.14" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32'", - "python_full_version >= '3.14' and sys_platform == 'emscripten'", - "python_full_version >= '3.14' and sys_platform == 'darwin'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", - "python_full_version == '3.13.*' and sys_platform == 'darwin'", - "python_full_version == '3.13.*' and sys_platform != 'darwin'", - "python_full_version == '3.12.*' and sys_platform == 'darwin'", - "python_full_version == '3.12.*' and sys_platform != 'darwin'", - "python_full_version == '3.11.*' and sys_platform == 'darwin'", - "python_full_version == '3.11.*' and sys_platform != 'darwin'", - "python_full_version < '3.11' and sys_platform == 'darwin'", - "python_full_version < '3.11' and sys_platform != 'darwin'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin'", ] sdist = { url = "https://files.pythonhosted.org/packages/6a/04/4a0812eb27c086cfd2e66e7ec9150f33e105912a9b7f8b335e3479f03a06/outlines_core-0.2.14.tar.gz", hash = "sha256:64808deed1591ca3029ff64346ceb974cd5d780c916ea82504951fe83523039e", size = 191539, upload-time = "2026-01-09T15:59:10.016Z" } wheels = [ @@ -6374,8 +7695,8 @@ name = "prometheus-fastapi-instrumentator" version = "8.0.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "prometheus-client" }, - { name = "starlette" }, + { name = "prometheus-client", marker = "python_full_version >= '3.11'" }, + { name = "starlette", marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/98/7a/fe49b7060e585e2cfa28cc1d13ebfe9c2904dcd80cf11071d5de0f622ff2/prometheus_fastapi_instrumentator-8.0.0.tar.gz", hash = "sha256:c31ed192db077e75467b28b2fe5055362517f53369b798cb8dac9ff85f3f1c38", size = 20357, upload-time = "2026-05-29T13:04:43.327Z" } wheels = [ @@ -6513,7 +7834,8 @@ name = "proto-plus" version = "1.27.2" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "protobuf" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/81/0d/94dfe80193e79d55258345901acd2917523d56e8381bc4dee7fd38e3868a/proto_plus-1.27.2.tar.gz", hash = "sha256:b2adde53adadf75737c44d3dcb0104fde65250dfc83ad59168b4aa3e574b6a24", size = 57204, upload-time = "2026-03-26T22:18:57.174Z" } wheels = [ @@ -6524,6 +7846,23 @@ wheels = [ name = "protobuf" version = "6.33.6" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] sdist = { url = "https://files.pythonhosted.org/packages/66/70/e908e9c5e52ef7c3a6c7902c9dfbb34c7e29c25d2f81ade3856445fd5c94/protobuf-6.33.6.tar.gz", hash = "sha256:a6768d25248312c297558af96a9f9c929e8c4cee0659cb07e780731095f38135", size = 444531, upload-time = "2026-03-18T19:05:00.988Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/fc/9f/2f509339e89cfa6f6a4c4ff50438db9ca488dec341f7e454adad60150b00/protobuf-6.33.6-cp310-abi3-win32.whl", hash = "sha256:7d29d9b65f8afef196f8334e80d6bc1d5d4adedb449971fefd3723824e6e77d3", size = 425739, upload-time = "2026-03-18T19:04:48.373Z" }, @@ -6535,6 +7874,43 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/c4/72/02445137af02769918a93807b2b7890047c32bfb9f90371cbc12688819eb/protobuf-6.33.6-py3-none-any.whl", hash = "sha256:77179e006c476e69bf8e8ce866640091ec42e1beb80b213c3900006ecfba6901", size = 170656, upload-time = "2026-03-18T19:04:59.826Z" }, ] +[[package]] +name = "protobuf" +version = "7.36.1" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +sdist = { url = "https://files.pythonhosted.org/packages/86/73/f66c748df06e7fe24e658eddd600d19c4b40bad836c97ce2d0ad9851fb6b/protobuf-7.36.1.tar.gz", hash = "sha256:d0f6470f0ce2b84e3feaea2d4b816378b37ba4d4aa08a274305373de93e2d524", size = 512499, upload-time = "2026-08-31T22:40:04.667Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/f7/6c/3a54a58f2948b0f485df9ecdd06590f15d0a7abf46a89d50c3de709ff4ff/protobuf-7.36.1-cp310-abi3-macosx_10_9_universal2.whl", hash = "sha256:3cf2ee25d006cee57294a1196ea43b37feb78e0dcd1e8af5c1aeddb777655aca", size = 456046, upload-time = "2026-08-31T22:39:56.865Z" }, + { url = "https://files.pythonhosted.org/packages/6e/08/9f9548793c771095245c0eeaf0c84b76b16a5ef26158043d456f551344a0/protobuf-7.36.1-cp310-abi3-manylinux2014_aarch64.whl", hash = "sha256:43d3d37b1eb24c113b9b7d02008cac44e423f00b611b7781ae998d7623972969", size = 344226, upload-time = "2026-08-31T22:39:58.263Z" }, + { url = "https://files.pythonhosted.org/packages/fe/51/1bdbd612fa3c51e42ea6f45d05e84dc748ddcd2663e1aa1e89a00a33facd/protobuf-7.36.1-cp310-abi3-manylinux2014_s390x.whl", hash = "sha256:39c518c05586c016d7874ff6079ee115bcec1ea5fbb1d177fbf7867ef4c67e44", size = 357229, upload-time = "2026-08-31T22:39:59.198Z" }, + { url = "https://files.pythonhosted.org/packages/22/df/c799fe7a05ef16ba853a59db01f3a2c5f7d0676469589ccc4874f76a2a88/protobuf-7.36.1-cp310-abi3-manylinux2014_x86_64.whl", hash = "sha256:97198b77e369a0abd8e262b8f6c7266c55ddb796a3a12c76d7b8881188ed83aa", size = 343228, upload-time = "2026-08-31T22:40:00.179Z" }, + { url = "https://files.pythonhosted.org/packages/3d/33/d4724ec5d86d496fe4108e220618aa0837ad3423a7af7ccdd9684ccc77c8/protobuf-7.36.1-cp310-abi3-win32.whl", hash = "sha256:0b53ce95272aad50ad25d7ff03373743209822e8ba42ea7fad27d2bee1547d00", size = 443002, upload-time = "2026-08-31T22:40:01.32Z" }, + { url = "https://files.pythonhosted.org/packages/db/37/155788a0d8daded960375af604202805308169f9b859419ea0aa370946e2/protobuf-7.36.1-cp310-abi3-win_amd64.whl", hash = "sha256:51139351435d9b43d88a55eaa49fb6f737fbb478fb0cbf2cf694d1a04a9d3363", size = 456518, upload-time = "2026-08-31T22:40:02.494Z" }, + { url = "https://files.pythonhosted.org/packages/39/ca/c47f91d3cab175b01fd8c4f0d80fdf8613be876cc616e66ad281a59c5ddf/protobuf-7.36.1-py3-none-any.whl", hash = "sha256:7d951e46b3f963d6c264c367c437921de9d5aedd9c3f9612b9077736b4e3ad5c", size = 179813, upload-time = "2026-08-31T22:40:03.54Z" }, +] + [[package]] name = "psutil" version = "7.2.2" @@ -6633,6 +8009,10 @@ wheels = [ name = "pyarrow" version = "23.0.1" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform != 'darwin'", +] sdist = { url = "https://files.pythonhosted.org/packages/88/22/134986a4cc224d593c1afde5494d18ff629393d74cc2eddb176669f234a4/pyarrow-23.0.1.tar.gz", hash = "sha256:b8c5873e33440b2bc2f4a79d2b47017a89c5a24116c055625e6f2ee50523f019", size = 1167336, upload-time = "2026-02-16T10:14:12.39Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/bc/a8/24e5dc6855f50a62936ceb004e6e9645e4219a8065f304145d7fb8a79d5d/pyarrow-23.0.1-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:3fab8f82571844eb3c460f90a75583801d14ca0cc32b1acc8c361650e006fd56", size = 34307390, upload-time = "2026-02-16T10:08:08.654Z" }, @@ -6686,6 +8066,93 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/50/f2/c0e76a0b451ffdf0cf788932e182758eb7558953f4f27f1aff8e2518b653/pyarrow-23.0.1-cp314-cp314t-win_amd64.whl", hash = "sha256:527e8d899f14bd15b740cd5a54ad56b7f98044955373a17179d5956ddb93d9ce", size = 28365807, upload-time = "2026-02-16T10:14:03.892Z" }, ] +[[package]] +name = "pyarrow" +version = "26.0.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +sdist = { url = "https://files.pythonhosted.org/packages/ec/34/17c34cb38e5d940e38f0f0d9fdfa0e8a506676409ea9b85aff7e3079f831/pyarrow-26.0.0.tar.gz", hash = "sha256:0cccd36e00ea3afeb52ded61f2721ce71f604853d70c45365c58324eb773d6ae", size = 1239433, upload-time = "2026-10-09T08:26:25.315Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/07/68/e0707097cee93be7f693e7e89495fabfeb8bf95ee30619063f8b30fffc29/pyarrow-26.0.0-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:fcdd1e04982637c6042337d3e24d472f938f01fdc502e2b994844b726d12c3f4", size = 36370896, upload-time = "2026-10-09T08:13:28.874Z" }, + { url = "https://files.pythonhosted.org/packages/5c/f0/591211c00612aef83236daff1620412b24aeb07c646de08c18a8a6c95a39/pyarrow-26.0.0-cp311-cp311-macosx_12_0_x86_64.whl", hash = "sha256:f800e9e722c145ccd18012d82a864cb21bfee4ba4ceffde77100d25eced511a9", size = 38709806, upload-time = "2026-10-09T08:13:33.417Z" }, + { url = "https://files.pythonhosted.org/packages/50/ea/9b035a9d1556e06e64ea86169d9a985d0fc092d427ac5edbb3af7183289c/pyarrow-26.0.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:7aa12ab8e236789b1ecd2d6ecaef036b4e63d675ddf1864a43c6799d18f2d028", size = 50885975, upload-time = "2026-10-09T08:13:37.737Z" }, + { url = "https://files.pythonhosted.org/packages/e1/81/8e685683897a6d3d5887c3e2fd24f3c14bc5d6d6bb3a2387484e665c580e/pyarrow-26.0.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:6e89dee53aaeb50505ed6152ea55bc7ddfd4f4df264f5427ea255288d8f0e580", size = 53904793, upload-time = "2026-10-09T08:13:42.984Z" }, + { url = "https://files.pythonhosted.org/packages/9a/ad/d474a0b1b00110f3a879aa5df654f857c81929a32b2a4222869240de5220/pyarrow-26.0.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:f1c1b4263fd13abbc339a16f2bf19f3a5cbf2a620853d812b1256f03c5342cb8", size = 54458010, upload-time = "2026-10-09T08:13:47.778Z" }, + { url = "https://files.pythonhosted.org/packages/d4/86/2c2861e905810c59fed4d98c85b994c21e8613730c5c3b436781d89110f2/pyarrow-26.0.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:ff1e816af7abff71f289242e109217036723ce36aca74ad6691e52d964a74afa", size = 57368406, upload-time = "2026-10-09T08:13:52.651Z" }, + { url = "https://files.pythonhosted.org/packages/0e/02/823e606633c15155bb965c7a0f3750c4f20dd47c4ab48213c7693df0e0ba/pyarrow-26.0.0-cp311-cp311-win_amd64.whl", hash = "sha256:13b0972a3dc71b642050d1bc72664a3916e14f59c943d8c1368154d6e4b0c2d5", size = 28522657, upload-time = "2026-10-09T08:13:56.513Z" }, + { url = "https://files.pythonhosted.org/packages/b3/60/6793778f2617cce469383dac0ba08c4f2401cf342df0c7b9ca53939d9b46/pyarrow-26.0.0-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:90ddaf7c625307ad52f31a9b25c34fe5e4897c7529ee3481135822b2b6842ff1", size = 36333953, upload-time = "2026-10-09T08:14:00.387Z" }, + { url = "https://files.pythonhosted.org/packages/db/81/f944cc63ce8a753e5fbff25de6d1d475ebd7fffdf9cf98c65130294fc896/pyarrow-26.0.0-cp312-cp312-macosx_12_0_x86_64.whl", hash = "sha256:ee341973f78a0b46e073d065e88e75026a9c584051e97f98a0d05d96c6bac7dd", size = 38688456, upload-time = "2026-10-09T08:14:04.344Z" }, + { url = "https://files.pythonhosted.org/packages/f5/2d/7e5c722fa5d5d9f3b75e62fe11694b34217664d4f05ac88031197166b277/pyarrow-26.0.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:01c863a18bd9c8412453dd0d92de6d0ee7b2b3d6fb079d9734a4b2a3c8bd4453", size = 50867603, upload-time = "2026-10-09T08:14:09.115Z" }, + { url = "https://files.pythonhosted.org/packages/88/e4/9cd356d906e71bd79b0c3fc5c9a54e01a0020dcf14c152ccfbcb503c7298/pyarrow-26.0.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:6a628922ba20705fa964ca73e4ef959c2fb2f14b9bbec5589a6a1e68e6257c85", size = 53931932, upload-time = "2026-10-09T08:14:24.051Z" }, + { url = "https://files.pythonhosted.org/packages/bb/e4/5bae3133b7fe04c24907a20f3bc1fba388cbbde659199e7b76445982047a/pyarrow-26.0.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:954d971b363b16ee41f89389a4053315dc71265f2ce5c2468eb0a910b1166268", size = 54444720, upload-time = "2026-10-09T08:14:31.214Z" }, + { url = "https://files.pythonhosted.org/packages/ba/b4/ee422493bb6dafdbef776cfe2c2a73106a1063a79bf4e78d1e5f51176885/pyarrow-26.0.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:5d5768d03426abe6526d5274adefa00abf00a7f81118c46e98b5a46390f5549e", size = 57388949, upload-time = "2026-10-09T08:14:38.964Z" }, + { url = "https://files.pythonhosted.org/packages/54/3c/1783aab1dac28e175dcf26dfc7123725efc474caecaed91e8a34cb89cad0/pyarrow-26.0.0-cp312-cp312-win_amd64.whl", hash = "sha256:cc903e1069e9dd5e9dcf780324c0112e27e051e422ecfaff574fb33ed65d9160", size = 28567581, upload-time = "2026-10-09T08:14:44.279Z" }, + { url = "https://files.pythonhosted.org/packages/4d/35/ca95493712af97c46a312945c8e9d16b21c5fe2f148be5466168d0290505/pyarrow-26.0.0-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:a6ca849f90cf73fe361f08a5762c783ead9671e4548c1f558cc637b54c9103f2", size = 36336700, upload-time = "2026-10-09T08:14:51.399Z" }, + { url = "https://files.pythonhosted.org/packages/69/ef/b1a675f79c9babfd4fcd99af62141d3c2d1a78a524e311b0c6b80110445a/pyarrow-26.0.0-cp313-cp313-macosx_12_0_x86_64.whl", hash = "sha256:c2ba350957076b1b3a22f549261dc3e9c67ca20816d8bd5f79d7b9c69be4c4c2", size = 38698502, upload-time = "2026-10-09T08:14:57.114Z" }, + { url = "https://files.pythonhosted.org/packages/3b/7c/cea852a832a327a8de797b3a68e5c25ce0f5aa1d20503807671bd90ec642/pyarrow-26.0.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:e3b190ba1d3d22a5a8758597f797111b77d433473744352a184a5ee0a42d672e", size = 50865064, upload-time = "2026-10-09T08:20:01.614Z" }, + { url = "https://files.pythonhosted.org/packages/4f/d6/e95834b29360092376fe4da9956ba41bb7b021869efe6ee9d4172d05cb15/pyarrow-26.0.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:240bd18a7487f8767616a948a69dd4e740a8bc36a1c9da49e4dc9a32c5c2faed", size = 53926722, upload-time = "2026-10-09T08:23:10.829Z" }, + { url = "https://files.pythonhosted.org/packages/e0/7f/98257444e2aea2e1fddceee3af3bd2077236d550428413f80393bd1f888d/pyarrow-26.0.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:2b5fcd69c0e1107b79e55839877db5a6ed04651b73fd6fec581d09e230bed5e4", size = 54443093, upload-time = "2026-10-09T08:23:16.971Z" }, + { url = "https://files.pythonhosted.org/packages/88/ca/dac99cfb25cfa62bf7194600cc99abc14a6bd2af50d7fdb7f15eeaf6e202/pyarrow-26.0.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:f7444ea6975c49a857c68f9bd8fa11acae96dede63d120ffb3bf0a603ea82516", size = 57381937, upload-time = "2026-10-09T08:23:24.95Z" }, + { url = "https://files.pythonhosted.org/packages/c0/ed/138d29fddaf803b90f4527e124bb6aaddc18aaf4a6c50fd0a5f577c94989/pyarrow-26.0.0-cp313-cp313-win_amd64.whl", hash = "sha256:3de30a7432b48b98b9decbd9e25a53bb9251d202c2e6c5a29a50869592ccb117", size = 28478571, upload-time = "2026-10-09T08:23:30.535Z" }, + { url = "https://files.pythonhosted.org/packages/8c/32/01858422a37f083911c2bb4d15cc32c5eeaa9d9b2bf5ddedee995a7146a6/pyarrow-26.0.0-cp314-cp314-macosx_12_0_arm64.whl", hash = "sha256:5780d487ff6c6ed7b42298609680d87fe0036e529a9dc2e1105364bce9697f50", size = 36378402, upload-time = "2026-10-09T08:23:36.537Z" }, + { url = "https://files.pythonhosted.org/packages/00/85/f6b5976c2878b752d0804d371684e0495a71de296b6dc6559e6fbaa4311a/pyarrow-26.0.0-cp314-cp314-macosx_12_0_x86_64.whl", hash = "sha256:a0e4e92eeb088f1d7c2c04d6c7de8434c75abb4b4ccf0bbcd045aa7164c68d93", size = 38733074, upload-time = "2026-10-09T08:23:42.873Z" }, + { url = "https://files.pythonhosted.org/packages/81/bc/c90fcbbcf893631e23dab1b0fb3fa29a508a8614326571b03c0894eda00b/pyarrow-26.0.0-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:eaf9e7cc7ab59f6c760232bbde18f64d559bbc50544841303bfb32be53533297", size = 50929201, upload-time = "2026-10-09T08:23:50.507Z" }, + { url = "https://files.pythonhosted.org/packages/ec/c1/0c1ff38ab7df1b2cf54cf0ad9f19a516c4e416c6c9b4c966cc2c9d587f77/pyarrow-26.0.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:ab6914db225d7f399652ae1f08588dfbc9efe617612715701e3d9d5cfa5ca19f", size = 53951865, upload-time = "2026-10-09T08:23:57.692Z" }, + { url = "https://files.pythonhosted.org/packages/9f/70/6a6b170496925472adad45a32528770fc8632db35fc60d4edd1e9ce1be0b/pyarrow-26.0.0-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:41dd3661ef40790a78870052ad7a58ad827b27c67a4511f06962eb9e9b74d19b", size = 54496388, upload-time = "2026-10-09T08:24:05.23Z" }, + { url = "https://files.pythonhosted.org/packages/a8/32/033ef9dba80976820190e292a10a5a23e9406572b76bbeb4d685d90e5c8d/pyarrow-26.0.0-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:6e949744dcfc2d379808f7013c5f9cafaf0f817656dff7d46c6931528dd1784b", size = 57411588, upload-time = "2026-10-09T08:24:12.043Z" }, + { url = "https://files.pythonhosted.org/packages/1e/ff/a74892c50aaf1f9f744a84493e08a2f99221e77c39d2d4a926de21a99edf/pyarrow-26.0.0-cp314-cp314-win_amd64.whl", hash = "sha256:4a5fa8dc70dd50808990ff36faf44088e357b353d86c7682dd92d4b78d4c97d5", size = 29237858, upload-time = "2026-10-09T08:24:58.106Z" }, + { url = "https://files.pythonhosted.org/packages/03/10/f0ee0976ef08a851a743c57608917ac9a47623f688b9ee0efe5429975ba1/pyarrow-26.0.0-cp314-cp314t-macosx_12_0_arm64.whl", hash = "sha256:e2a1856e9565fe2679863b372478c681806aebbf7d0a6e72f33e77f804e647d6", size = 36495870, upload-time = "2026-10-09T08:24:16.479Z" }, + { url = "https://files.pythonhosted.org/packages/27/ca/0bc431a509bf10b4472dbb94f4184752ecbbddeb7f467152dac0fdaed469/pyarrow-26.0.0-cp314-cp314t-macosx_12_0_x86_64.whl", hash = "sha256:4bcba83299cb2b8f8e443d36c6ba6269a5034431879015fb0719495df8a14de2", size = 38819754, upload-time = "2026-10-09T08:24:20.875Z" }, + { url = "https://files.pythonhosted.org/packages/61/59/2be41d26af7a07fb71581fb753cae396403ba1a2978355fd553929d44a9a/pyarrow-26.0.0-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:3a4d235876f14b4136b4d616ec42eb469ea0d6ead336cae631aa1dd29b21c962", size = 50933671, upload-time = "2026-10-09T08:24:27.199Z" }, + { url = "https://files.pythonhosted.org/packages/4b/cb/b6d5048cf3178be9678f5c9c60040199894b2f69c3439c87ced91fd24da9/pyarrow-26.0.0-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:210cc9b83888b87cdc8f793eebb264f22b20d0dedbedefc73b9687a7047b4747", size = 53906419, upload-time = "2026-10-09T08:24:33.536Z" }, + { url = "https://files.pythonhosted.org/packages/09/2b/23e30fbd776c81d18d134d2592eb60daca13e8a57ab087d0fa042f9d9f3d/pyarrow-26.0.0-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:ca77c43ca55bfc9a4eeb1f0cd5f093f08731b77c24cdba0829035f084959b0bb", size = 54527960, upload-time = "2026-10-09T08:24:41.292Z" }, + { url = "https://files.pythonhosted.org/packages/e2/23/fce251cd6b0546dfc181b00d5c8ef1c95a8c4cae83266bc3dfd5f719c62c/pyarrow-26.0.0-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:290a74c48e9491b436fd5edacfadf357943f82aa45c81110bd83a69aab33d1cf", size = 57388010, upload-time = "2026-10-09T08:24:48.186Z" }, + { url = "https://files.pythonhosted.org/packages/44/a5/0126fb0ef8d59bf257bdd68bb41623b72afc6e81790a0b4ac863a0f58861/pyarrow-26.0.0-cp314-cp314t-win_amd64.whl", hash = "sha256:515a10dae2a1d236bc9c9209d0317acb6746ea63cd4f98704904af7156d90ed1", size = 29406123, upload-time = "2026-10-09T08:24:53.387Z" }, + { url = "https://files.pythonhosted.org/packages/ed/66/8ada1b5165359d84b4b9b5384742304d1081da670f77d458fd9c9b8a2161/pyarrow-26.0.0-cp315-cp315-macosx_12_0_arm64.whl", hash = "sha256:e890816e5ee89c74a0f8b9379fe8b5ba83f46132b2a0bbb9b1c21359ec30dfda", size = 36373215, upload-time = "2026-10-09T08:25:03.067Z" }, + { url = "https://files.pythonhosted.org/packages/c4/83/74f10c3d803a6834b2acab21847724d4bdbc74d246eb17321432844707f3/pyarrow-26.0.0-cp315-cp315-macosx_12_0_x86_64.whl", hash = "sha256:9db18a9dc0af52135c9eac549d80a7a882696efbe5406cf882b044525d4ecc2e", size = 38730866, upload-time = "2026-10-09T08:25:07.924Z" }, + { url = "https://files.pythonhosted.org/packages/e2/5a/ea2fa2163b1bd8ff73efd39c4060be63fd6ddec03e7887a471acd1e042a4/pyarrow-26.0.0-cp315-cp315-manylinux_2_28_aarch64.whl", hash = "sha256:734312d3d99088d9ec28c5b17bad40389bd8373a1afc10acb60b83fd217af087", size = 50924443, upload-time = "2026-10-09T08:25:13.864Z" }, + { url = "https://files.pythonhosted.org/packages/78/80/8c47b6cf8cfd42826df65193eff026c1cc81fa6cb213a3c3f5d203e6f67a/pyarrow-26.0.0-cp315-cp315-manylinux_2_28_x86_64.whl", hash = "sha256:24f892fdf1ae1942d69d3f7742e2f49960ec95277cfb1a70b8a1d91f4a96d935", size = 53948540, upload-time = "2026-10-09T08:25:19.305Z" }, + { url = "https://files.pythonhosted.org/packages/69/1f/3a506a76d944ec5c5e4b7f01d8d0446b392a6fb384de627a12e503f616b4/pyarrow-26.0.0-cp315-cp315-musllinux_1_2_aarch64.whl", hash = "sha256:879331ddea2a26479fa18fade71e6facf684a6cf19f67daec3775c871569e8e5", size = 54494863, upload-time = "2026-10-09T08:25:24.517Z" }, + { url = "https://files.pythonhosted.org/packages/3d/50/08c4bb04d651788d2eaca78065743f4f6ded974d4ef96ae3c473993e9d0c/pyarrow-26.0.0-cp315-cp315-musllinux_1_2_x86_64.whl", hash = "sha256:5b827650e874f1f9f9392524ea3e9e3e8a245de5ba64acca1f81ab188090afb9", size = 57409877, upload-time = "2026-10-09T08:25:31.157Z" }, + { url = "https://files.pythonhosted.org/packages/d4/f3/c64781fbd7b6d3c07993b698c14944d0d195f07e800fa931c486ae6ab36a/pyarrow-26.0.0-cp315-cp315-win_amd64.whl", hash = "sha256:8e8e28c464552b5ca03e30d4504168c4425ce383884f8611b00e972f9fd933fc", size = 29236658, upload-time = "2026-10-09T08:26:22.607Z" }, + { url = "https://files.pythonhosted.org/packages/06/55/2ee3729daea999f19f061f03898d4895a242c4cd94f26e1324e5fdfbfe10/pyarrow-26.0.0-cp315-cp315t-macosx_12_0_arm64.whl", hash = "sha256:ce28748cbeb0f29c3ce9603782979c7117580fc76f16aa3ca448b38a22281adb", size = 36489011, upload-time = "2026-10-09T08:25:37.64Z" }, + { url = "https://files.pythonhosted.org/packages/6a/7d/3eb17f601f2bf13eda5f2ed28956379ca628b4dda97619cbb1cb1721622d/pyarrow-26.0.0-cp315-cp315t-macosx_12_0_x86_64.whl", hash = "sha256:106bb9290fc6fd9a84138a9440038ef184bac86463543c5ff099229cb30d996c", size = 38808480, upload-time = "2026-10-09T08:25:43.579Z" }, + { url = "https://files.pythonhosted.org/packages/0e/e3/f0047360b0f4bfc031b256dc0aec3837a61f245b2fb70f8363438e2db665/pyarrow-26.0.0-cp315-cp315t-manylinux_2_28_aarch64.whl", hash = "sha256:2e4a413046eba9896e632925066c74095182200ba32e19ff0166bf64d2f936ac", size = 50923273, upload-time = "2026-10-09T08:25:51.445Z" }, + { url = "https://files.pythonhosted.org/packages/38/d9/56d9fb91210407df31cbeb9b91138601c88c7c8fb5f6bf773b20d65509bf/pyarrow-26.0.0-cp315-cp315t-manylinux_2_28_x86_64.whl", hash = "sha256:d58798c4d8d629700058e9afc1e16b9801023f3ce4dc1c92d945e79b5ffe4e98", size = 53900905, upload-time = "2026-10-09T08:25:59.554Z" }, + { url = "https://files.pythonhosted.org/packages/cf/40/8e8a7e9e027c731520c7eb179dd00a153b76ebf0bc11d213c6c8f8502851/pyarrow-26.0.0-cp315-cp315t-musllinux_1_2_aarch64.whl", hash = "sha256:645917e976671debabf854abab6e2b75c571ca4f82adc33a2d338697f7c27d93", size = 54518345, upload-time = "2026-10-09T08:26:07.125Z" }, + { url = "https://files.pythonhosted.org/packages/be/89/1e768a3fdb88d34e708ad2dc00dbf8e4e30290784eb84198d59308963bea/pyarrow-26.0.0-cp315-cp315t-musllinux_1_2_x86_64.whl", hash = "sha256:7c3fda041e7078802589cf257750323ee3d0cd1e56e53a9b20ec845697fb3d28", size = 57379403, upload-time = "2026-10-09T08:26:13.624Z" }, + { url = "https://files.pythonhosted.org/packages/96/be/7b81a44d6a8e70581dcc1d6f01541f9000a973b1e5d75394aec91e7b179a/pyarrow-26.0.0-cp315-cp315t-win_amd64.whl", hash = "sha256:68cd662e9e2b00876a131950cf32336ace2d0865e1f9418763e3d3be8481dfa4", size = 29389953, upload-time = "2026-10-09T08:26:18.277Z" }, +] + [[package]] name = "pyasn1" version = "0.6.4" @@ -6947,8 +8414,10 @@ name = "pydantic" version = "2.12.5" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -6972,7 +8441,7 @@ wheels = [ [package.optional-dependencies] email = [ - { name = "email-validator", marker = "python_full_version < '3.12'" }, + { name = "email-validator", marker = "python_full_version == '3.11.*'" }, ] [[package]] @@ -6980,18 +8449,26 @@ name = "pydantic" version = "2.13.4" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -7017,8 +8494,10 @@ name = "pydantic-core" version = "2.41.5" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -7148,18 +8627,26 @@ name = "pydantic-core" version = "2.46.4" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -7281,10 +8768,10 @@ name = "pydantic-extra-types" version = "2.11.2" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-sglang') or (python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/41/d3/3be31542180c0300b6860129ff1e3a428f3ef580727616ce22462626129b/pydantic_extra_types-2.11.2.tar.gz", hash = "sha256:3a2b83b61fe920925688e7838b59caa90a45637d1dbba2b1364b8d1f7ff72a0a", size = 203929, upload-time = "2026-04-05T20:50:51.556Z" } wheels = [ @@ -7293,7 +8780,7 @@ wheels = [ [package.optional-dependencies] pycountry = [ - { name = "pycountry" }, + { name = "pycountry", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, ] [[package]] @@ -7301,25 +8788,16 @@ name = "pydantic-settings" version = "2.14.2" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "python-dotenv" }, - { name = "typing-inspection" }, + { name = "python-dotenv", marker = "python_full_version >= '3.11'" }, + { name = "typing-inspection", marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/5c/b5/8f48e906c3e0205276e8bd8cb7512217a87b2685304d64be27cad5b3019f/pydantic_settings-2.14.2.tar.gz", hash = "sha256:c19dd64b19097f1de80184f0cc7b0272a13ae6e170cbf240a3e27e381ed14a5f", size = 237700, upload-time = "2026-06-19T13:44:56.324Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/77/c1/6e422f34e569cf8e18df68d1939c81c099d2b61e4f7d9621c8a77560799c/pydantic_settings-2.14.2-py3-none-any.whl", hash = "sha256:a20c97b37910b6550d5ea50fbcc2d4187defe58cd57070b73863d069419c9440", size = 61715, upload-time = "2026-06-19T13:44:55.02Z" }, ] -[[package]] -name = "pyelftools" -version = "0.33" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/a3/11/767522582afab1b884d277de0e6e011640cb9d7292a38694b4b1a1df1ae8/pyelftools-0.33.tar.gz", hash = "sha256:660d82dcbeb8e83d1702bd97f223f761625da06111c0cc988eac6b8ab0c1b61f", size = 15068655, upload-time = "2026-05-29T12:56:22.553Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/46/2a/f9697576603dae937727827505a6126a066affb227034e77e6f9068910da/pyelftools-0.33-py3-none-any.whl", hash = "sha256:f215ad5f47d3f1373a21496a6c9e0707c622840d0622f23ff7ce08678b020036", size = 201178, upload-time = "2026-05-29T12:56:20.587Z" }, -] - [[package]] name = "pygments" version = "2.20.0" @@ -7333,9 +8811,6 @@ wheels = [ name = "pyjwt" version = "2.15.0" source = { registry = "https://pypi.org/simple" } -dependencies = [ - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, -] sdist = { url = "https://files.pythonhosted.org/packages/02/a5/5197bfd06417837ac079921c66fa6393f1dea3557272a263cebfef69e432/pyjwt-2.15.0.tar.gz", hash = "sha256:b11c5f9791d7bf51c2b39a81ed669f6b2dbbd669df2942f6c60167e9e3d1abe4", size = 120513, upload-time = "2026-09-23T16:56:00.689Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/e8/55/40e45bf052ee8ee12a4dfd785519660f8effa7b065442b91646ec6828619/pyjwt-2.15.0-py3-none-any.whl", hash = "sha256:7a3742debf6b879e912dbb9819ceec1594be812452b78c5f2e2dfc56564954f8", size = 33680, upload-time = "2026-09-23T16:55:59.241Z" }, @@ -7343,7 +8818,7 @@ wheels = [ [package.optional-dependencies] crypto = [ - { name = "cryptography" }, + { name = "cryptography", marker = "python_full_version >= '3.11'" }, ] [[package]] @@ -7413,6 +8888,31 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/03/bd/76eecf37a469a2e19c82c85b20402cbeece261ef9d9286189502f39e3566/pymoo-0.6.1.6-cp314-cp314-win_amd64.whl", hash = "sha256:56ecb6f9f5ac0559829e183a23ea3672eae8c1eb9d745305bef98da8d1614d28", size = 1851874, upload-time = "2025-11-25T03:18:28.969Z" }, ] +[[package]] +name = "pynvvideocodec" +version = "2.0.4" +source = { registry = "https://pypi.org/simple" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/c0/9d/bcde031d2b7ee86336459236729dbb7a8a58d6fef0b0d78f99fe8929db13/pynvvideocodec-2.0.4-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:c7a133a8088ff9152f7953ff63c5cbc27801f0313d1e577c5f3e65da67abf570", size = 28632415, upload-time = "2026-07-08T04:25:07.008Z" }, + { url = "https://files.pythonhosted.org/packages/8d/a9/b80e3b2cb324f4fa684f66748f1fa1f14e8e1687309408cf3e1e8c1c2a81/pynvvideocodec-2.0.4-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:7af1d9ab37f7415168c23be00539094e7b21774661af2a0a2b9bb574670bc75e", size = 43180627, upload-time = "2026-05-27T04:00:41.809Z" }, + { url = "https://files.pythonhosted.org/packages/92/d7/a57b8672ea2fb5edfb53a61b2478859d52c593d40c1e061b558c31a743c4/pynvvideocodec-2.0.4-cp310-cp310-manylinux_2_34_aarch64.whl", hash = "sha256:8d6d3d63317c452bdd18167f622abb392bd982ae65293673a169e3f253c1d1ed", size = 35754785, upload-time = "2026-05-27T04:01:15.377Z" }, + { url = "https://files.pythonhosted.org/packages/72/bf/e181180f9e9a60dac64640f8eeaf5635184c6aa11663f7855339c440f3a5/pynvvideocodec-2.0.4-cp310-cp310-win_amd64.whl", hash = "sha256:f678e5a6903498f291cffaba5d47b023c67bf18172049186044c4466492ad443", size = 25687790, upload-time = "2026-05-27T04:02:05.621Z" }, + { url = "https://files.pythonhosted.org/packages/ee/ce/4559ca81f39b14cc121ed284afc017fe36c3aa40e72ef12170b75d7d2113/pynvvideocodec-2.0.4-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:4dba42331f6d319087d05359787c7c542483fab22107c07b983cad928c1e3cfe", size = 28632422, upload-time = "2026-07-08T04:25:29.591Z" }, + { url = "https://files.pythonhosted.org/packages/2e/86/8766b11b0884fe9c0530870e99c68af1ce68568ebe0652b636caaa9ea50a/pynvvideocodec-2.0.4-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:bd3779ff73ad703393c0a19c3650f269ca25e71902c24efa0719ed4a58cd9390", size = 43180651, upload-time = "2026-05-27T04:02:38.117Z" }, + { url = "https://files.pythonhosted.org/packages/a9/aa/dd826d41581aa6b11a4c922752a89a0dfbfe5f9095de11f8bb10bde46c1d/pynvvideocodec-2.0.4-cp311-cp311-manylinux_2_34_aarch64.whl", hash = "sha256:f809fb18929ac2af042835f10c7679b1c86db9817776f22bc7467907c5c3d918", size = 35755024, upload-time = "2026-05-27T04:03:02.004Z" }, + { url = "https://files.pythonhosted.org/packages/cc/df/c5b516bb16c42c312d3564999f25487607ed105f656ef7fbeb6b13ff3c6e/pynvvideocodec-2.0.4-cp311-cp311-win_amd64.whl", hash = "sha256:fc299a14e61832850be91f8441669ada6fc270903ad1c50acde8e9353590fd3a", size = 25688228, upload-time = "2026-05-27T04:03:29.487Z" }, + { url = "https://files.pythonhosted.org/packages/38/1c/78f6fdf85133157a6a3405eab5ef4c2bc8048194dbda1c91bb9b8645bb36/pynvvideocodec-2.0.4-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:bad9e25f494abdcfa8f9dffa33a840509eda3ffcdf6e7cf6465d73be307c0c82", size = 28630316, upload-time = "2026-07-08T04:25:54.596Z" }, + { url = "https://files.pythonhosted.org/packages/ca/49/98da271686e00676f41b1197ba5431ddc341b96d8efb68ea9d68e2b0d870/pynvvideocodec-2.0.4-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:b59cec7a1a3f78fad13fead78cad8b6d9686827f9ff4477080245457675a01d0", size = 43176147, upload-time = "2026-05-27T04:04:08.297Z" }, + { url = "https://files.pythonhosted.org/packages/42/80/7b13c12fd5f3243b01190130ce098a44ddf62e030e6ed712911cbfe40311/pynvvideocodec-2.0.4-cp312-cp312-manylinux_2_34_aarch64.whl", hash = "sha256:a0daa28b09705806c8c6b26326df217c45e60c0a12a673ea3ea6ee5e2e7193b0", size = 35754893, upload-time = "2026-05-27T04:04:43.866Z" }, + { url = "https://files.pythonhosted.org/packages/76/50/eb1571ab1cee8ebb8a7bdfc355078beebe4b2bb2e5c6ad5d0e18ab8585db/pynvvideocodec-2.0.4-cp312-cp312-win_amd64.whl", hash = "sha256:46e2adb82dc6ac333d3535cc76e4e25c7e8d80dd272b1aba0c28702b861d5261", size = 25692590, upload-time = "2026-05-27T04:05:17.164Z" }, + { url = "https://files.pythonhosted.org/packages/46/d6/8475720d4f0f8fc9e852e0092b308643b71f24d9495ba3bd03c38b16c28d/pynvvideocodec-2.0.4-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:fcb06ef6ef24ee33b8f34950696a4e01636f2d8cdb96c407cd692931d049ac3d", size = 43176124, upload-time = "2026-05-27T04:06:26.415Z" }, + { url = "https://files.pythonhosted.org/packages/3d/28/33d2a4d69b48823801c0b02b1bfbeac04cd9d429ee698d248e7ba946c516/pynvvideocodec-2.0.4-cp313-cp313-manylinux_2_34_aarch64.whl", hash = "sha256:51724c6a0e3623c092cccdf93c8b09cced6881f3d0c76653f6ffbec0371f29cd", size = 35754919, upload-time = "2026-05-27T04:07:09.962Z" }, + { url = "https://files.pythonhosted.org/packages/49/8e/8c08bbffc9021db762b6ac103ce544db4e517f9816118a35db9f0c7d2a9a/pynvvideocodec-2.0.4-cp313-cp313-win_amd64.whl", hash = "sha256:8e704a2b553a35cc2de10543a232e7feddc473f3e5478996595236e3194efc5d", size = 25692569, upload-time = "2026-05-27T04:07:38.663Z" }, + { url = "https://files.pythonhosted.org/packages/a7/ce/3062bf51bc0d98a344b9ed229a40535f6f1d309fc9ffaf34719b67ed3e21/pynvvideocodec-2.0.4-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:d9ec06f47bca7b20a6e8234afaf596699c57a477be292613d676802e7e808ed2", size = 43174764, upload-time = "2026-05-27T04:08:19.02Z" }, + { url = "https://files.pythonhosted.org/packages/be/de/27eedeaa3ee5ec3dc2e4b1a7ac2b3d15ebb04cbfa768ca03d159a8328612/pynvvideocodec-2.0.4-cp314-cp314-manylinux_2_34_aarch64.whl", hash = "sha256:834fdbef7f3fc79285b5c2a88d1f1f7cd13543a9a7f2a2786141b181d1daaac9", size = 35755640, upload-time = "2026-05-27T04:08:45.728Z" }, + { url = "https://files.pythonhosted.org/packages/f1/9a/ff638f3203b3e49718760a68e7f573747d3d814b817d949501243405c9d0/pynvvideocodec-2.0.4-cp314-cp314-win_amd64.whl", hash = "sha256:fbe730967d5402ffca520b12fa21725cbd22a6d2c9fae882ea1d95085a400fd9", size = 25694550, upload-time = "2026-05-27T04:09:11.735Z" }, +] + [[package]] name = "pyparsing" version = "3.3.2" @@ -7462,7 +8962,8 @@ dependencies = [ { name = "numpy" }, { name = "opt-einsum" }, { name = "pyro-api" }, - { name = "torch" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11' or extra == 'extra-7-vec-inf-sglang'" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "tqdm" }, ] sdist = { url = "https://files.pythonhosted.org/packages/4c/2e/3bcba8688d58f8dc954cef6831c19d52b6017b035d783685d67cd99fa351/pyro_ppl-1.9.1.tar.gz", hash = "sha256:5e1596de276c038a3f77d2580a90d0a97126e0104900444a088eee620bb0d65e", size = 570861, upload-time = "2024-06-02T00:37:39.688Z" } @@ -7693,7 +9194,7 @@ name = "pyzmq" version = "27.1.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "cffi", marker = "implementation_name == 'pypy'" }, + { name = "cffi", marker = "(python_full_version >= '3.11' and implementation_name == 'pypy' and extra == 'extra-7-vec-inf-vllm') or (implementation_name == 'pypy' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/04/0b/3c9baedbdf613ecaa7aa07027780b8867f57b6293b6ee50de316c9f3222b/pyzmq-27.1.0.tar.gz", hash = "sha256:ac0765e3d44455adb6ddbf4417dcce460fc40a05978c08efdf2948072f6db540", size = 281750, upload-time = "2025-09-08T23:10:18.157Z" } wheels = [ @@ -7765,12 +9266,23 @@ wheels = [ name = "quack-kernels" version = "0.4.1" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and sys_platform != 'darwin'", + "python_full_version < '3.11' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform != 'darwin'", +] dependencies = [ { name = "apache-tvm-ffi", version = "0.1.9", source = { registry = "https://pypi.org/simple" } }, { name = "einops" }, - { name = "nvidia-cutlass-dsl", version = "4.5.1", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, - { name = "nvidia-cutlass-dsl", version = "4.5.2", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-vllm'" }, - { name = "torch" }, + { name = "nvidia-cutlass-dsl", version = "4.5.1", source = { registry = "https://pypi.org/simple" } }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" } }, { name = "torch-c-dlpack-ext" }, ] sdist = { url = "https://files.pythonhosted.org/packages/2a/58/58b82e91b236539f424ff5681e7095b1f2860ddfb7778fe0be14d8fb58de/quack_kernels-0.4.1.tar.gz", hash = "sha256:9d7d6ba412bc0c8a9b1331c52a73db76280adb9dc2f2750df4851ddabef1466b", size = 274766, upload-time = "2026-04-30T14:37:55.65Z" } @@ -7778,18 +9290,58 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/38/e4/a6c3bbbe3d4242fa412454b8e8069a079e500be331aecf8f2aa666164e9c/quack_kernels-0.4.1-py3-none-any.whl", hash = "sha256:c1c8df2935bf5156ec47d2c5384ac08b411fd0ee702d80ae916dbf6d6f5ae813", size = 260827, upload-time = "2026-04-30T14:37:54.584Z" }, ] +[[package]] +name = "quack-kernels" +version = "0.6.5" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin'", +] +dependencies = [ + { name = "apache-tvm-ffi", version = "0.1.11", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "einops", marker = "python_full_version >= '3.11'" }, + { name = "nvidia-cutlass-dsl", version = "4.8.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "torch-c-dlpack-ext", marker = "python_full_version >= '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/5f/18/6c2a20d00dff254f154fb3b951f4e5a0377a60122c17385e5f08d5f5b5e1/quack_kernels-0.6.5.tar.gz", hash = "sha256:dd9405a4a6fa35fc3a5a18fe35b46a14b0bf7e3ad9fe2451903e355047ac471d", size = 846876, upload-time = "2026-09-11T05:08:20.621Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/73/d8/f5e56c38b286308d115a171e3c5c44f24e9d03980c3cfd22588df40261fa/quack_kernels-0.6.5-py3-none-any.whl", hash = "sha256:df1ddd31366c82eb9692fe087c671209189dcbde53c909c81a865d4ee66b158d", size = 729615, upload-time = "2026-09-11T05:08:19.306Z" }, +] + [[package]] name = "ray" version = "2.56.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "click" }, + { name = "click", version = "8.3.2", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "filelock", version = "4.1.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "filelock", version = "4.2.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "jsonschema" }, { name = "msgpack" }, { name = "packaging" }, - { name = "protobuf" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pyyaml" }, { name = "requests" }, ] @@ -7976,7 +9528,8 @@ source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "certifi" }, { name = "charset-normalizer" }, - { name = "idna" }, + { name = "idna", version = "3.15", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "idna", version = "3.20", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "urllib3" }, ] sdist = { url = "https://files.pythonhosted.org/packages/5f/a4/98b9c7c6428a668bf7e42ebb7c79d576a1c3c1e3ae2d47e674b468388871/requests-2.33.1.tar.gz", hash = "sha256:18817f8c57c6263968bc123d237e3b8b08ac046f5456bd1e307ee8f4250d3517", size = 134120, upload-time = "2026-03-30T16:09:15.531Z" } @@ -8002,10 +9555,9 @@ name = "rich-toolkit" version = "0.19.7" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "click" }, - { name = "rich" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "rich", marker = "python_full_version >= '3.11'" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/42/ba/dae9e3096651042754da419a4042bc1c75e07d615f9b15066d738838e4df/rich_toolkit-0.19.7.tar.gz", hash = "sha256:133c0915872da91d4c25d85342d5ec1dfacc69b63448af1a08a0d4b4f23ef46e", size = 195877, upload-time = "2026-02-24T16:06:20.555Z" } wheels = [ @@ -8284,6 +9836,13 @@ wheels = [ name = "safetensors" version = "0.7.0" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] sdist = { url = "https://files.pythonhosted.org/packages/29/9c/6e74567782559a63bd040a236edca26fd71bc7ba88de2ef35d75df3bca5e/safetensors-0.7.0.tar.gz", hash = "sha256:07663963b67e8bd9f0b8ad15bb9163606cd27cc5a1b96235a50d8369803b96b0", size = 200878, upload-time = "2025-11-19T15:18:43.199Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/fa/47/aef6c06649039accf914afef490268e1067ed82be62bcfa5b7e886ad15e8/safetensors-0.7.0-cp38-abi3-macosx_10_12_x86_64.whl", hash = "sha256:c82f4d474cf725255d9e6acf17252991c3c8aac038d6ef363a4bf8be2f6db517", size = 467781, upload-time = "2025-11-19T15:18:35.84Z" }, @@ -8306,6 +9865,62 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/58/5b/632a58724221ef03d78ab65062e82a1010e1bef8e8e0b9d7c6d7b8044841/safetensors-0.7.0-pp310-pypy310_pp73-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:473b32699f4200e69801bf5abf93f1a4ecd432a70984df164fc22ccf39c4a6f3", size = 531885, upload-time = "2025-11-19T15:18:27.146Z" }, ] +[[package]] +name = "safetensors" +version = "0.8.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +sdist = { url = "https://files.pythonhosted.org/packages/45/06/f955dbbb1859e3bd23c8ac6141af5106e7ad5fedec4a3a6e3d60f94b7001/safetensors-0.8.0.tar.gz", hash = "sha256:fabaf3e0f18a6618d9b36560682562157f77c2b71fcffc7b432be2baed9d753d", size = 325846, upload-time = "2026-06-09T07:52:25.563Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/39/a0/f718cda65b05407d228f97602cf60dca269c979867aa5beb25410de26cd3/safetensors-0.8.0-cp310-abi3-macosx_10_12_x86_64.whl", hash = "sha256:c554f85858e05226d3c2828e32395e677434685d6d94594a41643361c5e837f0", size = 473568, upload-time = "2026-06-09T07:52:18.829Z" }, + { url = "https://files.pythonhosted.org/packages/f5/b1/fa7c600e7dceae12e9606c7578cbc9ff1e1ed55844883ee5c92205e86226/safetensors-0.8.0-cp310-abi3-macosx_11_0_arm64.whl", hash = "sha256:c80201d22cbf405b80647a60ada77bba06c8fba2da2743ba1e89cdcc39a81f25", size = 484562, upload-time = "2026-06-09T07:52:17.518Z" }, + { url = "https://files.pythonhosted.org/packages/09/7d/65a7de0af421317bb36a067241e4235fff194eed60b961ed6d3f59a3fc60/safetensors-0.8.0-cp310-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7a46e5ff292c356d6991e60942ba7f79817682d3a2cef0702136448cb9c4d235", size = 502844, upload-time = "2026-06-09T07:52:07.624Z" }, + { url = "https://files.pythonhosted.org/packages/91/4f/3175c9d75634e0e0dda0082794193521035edd7c70a6f212bf33ca06ddf4/safetensors-0.8.0-cp310-abi3-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:4124502b78f03534117c848f87a39b8f31e577b15eff423bf8bfb95f2a8c30d0", size = 511823, upload-time = "2026-06-09T07:52:09.565Z" }, + { url = "https://files.pythonhosted.org/packages/20/87/846c289e7aa2299eff406335717cf43ce8777194ece8aad75772e0411615/safetensors-0.8.0-cp310-abi3-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:7bc0a787ba8a35be368ee3574edfa2b1ad389eebd0a72e482ae275490e3f6c98", size = 633461, upload-time = "2026-06-09T07:52:11.128Z" }, + { url = "https://files.pythonhosted.org/packages/76/22/8d64d9df2c45d5ded401df889d0ad90882804ca172d79ec4f0df8f727fe0/safetensors-0.8.0-cp310-abi3-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:040070828e36dc8e122178bbbd5830ff9e97920affb84cbe0f46442497bed358", size = 545148, upload-time = "2026-06-09T07:52:13.603Z" }, + { url = "https://files.pythonhosted.org/packages/28/50/f203ff3a3ddfe19308efc83c5a3a29ed02bf786732ec35e68bf9162f3365/safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fd6f3f93c9a0a7cc2788ee63fb763353d4bd2e89b0751bc78fcf7dda00bea774", size = 516040, upload-time = "2026-06-09T07:52:16.29Z" }, + { url = "https://files.pythonhosted.org/packages/46/fb/cdaed17ceb2948784fd9c36b6fd3e951b608547cea81a48e8ee6f8cfdfcb/safetensors-0.8.0-cp310-abi3-manylinux_2_31_riscv64.whl", hash = "sha256:fcdd41ec4628fee5799f807c73c353629130fbd942aa23d83c623dd6c9d52d78", size = 513832, upload-time = "2026-06-09T07:52:12.37Z" }, + { url = "https://files.pythonhosted.org/packages/0d/49/1e15de264dcc3b77943d2d0c56a95809956883b1c2d6d585c792523f180b/safetensors-0.8.0-cp310-abi3-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:8e9f537aa183a38ace122d27303dcd986b26bd2a7591f9181d7f0c396f4677ca", size = 559930, upload-time = "2026-06-09T07:52:14.743Z" }, + { url = "https://files.pythonhosted.org/packages/2a/43/bf38443278eab4b1be1fce2931e2b012ad9cb7df52ada751d0aab8f7659a/safetensors-0.8.0-cp310-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:87eec7ffed2b809f05a398a8becb7d013f19f7837cd15d9748580d6cf30dbaf4", size = 678670, upload-time = "2026-06-09T07:52:20.032Z" }, + { url = "https://files.pythonhosted.org/packages/72/e3/68cd3fa5b48488e84add63e04cb12f3bc28ae4638c06d4508c6e88823d0e/safetensors-0.8.0-cp310-abi3-musllinux_1_2_armv7l.whl", hash = "sha256:4a95ae2b05d7726d751da4ebf626a2ca782b706e101bd894c95bc2450b1cffcc", size = 786679, upload-time = "2026-06-09T07:52:21.322Z" }, + { url = "https://files.pythonhosted.org/packages/29/4b/1c19c509d56e01f4fbb3d0a2e597450f6cc04d1d56cf52defb0a62dfd715/safetensors-0.8.0-cp310-abi3-musllinux_1_2_i686.whl", hash = "sha256:3ae091f16662658bdc019a4ff6cb4c085bb7d725eb5978b183ffd265863b6d2d", size = 765683, upload-time = "2026-06-09T07:52:22.594Z" }, + { url = "https://files.pythonhosted.org/packages/27/43/41c1621732edd934d868a00d1b891584c892a7b62a9aab82ea5a0a5623ee/safetensors-0.8.0-cp310-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:8e080062fcde23be189565e1c3305d16751a218ecf9412c8601e64204eb6f846", size = 722361, upload-time = "2026-06-09T07:52:23.924Z" }, + { url = "https://files.pythonhosted.org/packages/8e/3f/73ccf82579412b4a71c4ca673f10b5f1f888d7cf5af7fe24f27d30307be4/safetensors-0.8.0-cp310-abi3-win32.whl", hash = "sha256:2ddf52eac562eda224f99acfa7889d02968c1fd59a5b011ae7d8137c37e9c02d", size = 342401, upload-time = "2026-06-09T07:52:28.895Z" }, + { url = "https://files.pythonhosted.org/packages/1b/6d/3fba214c1e5e0f69991677ec3bc17023f0421776975e1de0c682dca475e2/safetensors-0.8.0-cp310-abi3-win_amd64.whl", hash = "sha256:096ec1a98435df7beb08853bb5aa9081a84f23d0adc67ed1a0a10550f608373f", size = 355540, upload-time = "2026-06-09T07:52:27.832Z" }, + { url = "https://files.pythonhosted.org/packages/8d/fc/7eedc3510d97878876e32774eebbeb61c43f148a96e915c84229a3e967aa/safetensors-0.8.0-cp310-abi3-win_arm64.whl", hash = "sha256:f7838e5135a406ad3e02efdcb8cf2e5397d368b0154537c4fec682dbc544d452", size = 340500, upload-time = "2026-06-09T07:52:26.745Z" }, +] + [[package]] name = "scikit-learn" version = "1.7.2" @@ -8362,23 +9977,33 @@ name = "scikit-learn" version = "1.8.0" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -8497,23 +10122,33 @@ name = "scipy" version = "1.17.1" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -8590,9 +10225,9 @@ name = "seaborn" version = "0.13.2" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "matplotlib" }, - { name = "numpy" }, - { name = "pandas" }, + { name = "matplotlib", marker = "python_full_version >= '3.11'" }, + { name = "numpy", marker = "python_full_version >= '3.11'" }, + { name = "pandas", marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/86/59/a451d7420a77ab0b98f7affa3a1d78a313d2f7281a57afb1a34bae8ab412/seaborn-0.13.2.tar.gz", hash = "sha256:93e60a40988f4d65e9f4885df477e2fdaff6b73a9ded434c1ab356dd57eefff7", size = 1457696, upload-time = "2024-01-25T13:21:52.551Z" } wheels = [ @@ -8668,8 +10303,8 @@ name = "sentry-sdk" version = "2.57.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "certifi" }, - { name = "urllib3" }, + { name = "certifi", marker = "python_full_version < '3.11' or extra == 'extra-7-vec-inf-vllm'" }, + { name = "urllib3", marker = "python_full_version < '3.11' or extra == 'extra-7-vec-inf-vllm'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/4f/87/46c0406d8b5ddd026f73adaf5ab75ce144219c41a4830b52df4b9ab55f7f/sentry_sdk-2.57.0.tar.gz", hash = "sha256:4be8d1e71c32fb27f79c577a337ac8912137bba4bcbc64a4ec1da4d6d8dc5199", size = 435288, upload-time = "2026-03-31T09:39:29.264Z" } wheels = [ @@ -8801,15 +10436,17 @@ dependencies = [ { name = "av", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'arm64' and sys_platform == 'linux') or (platform_machine == 'armv7l' and sys_platform == 'linux')" }, { name = "blobfile" }, { name = "build" }, - { name = "compressed-tensors" }, + { name = "compressed-tensors", version = "0.15.0.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "compressed-tensors", version = "0.17.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "cuda-python" }, - { name = "datasets" }, + { name = "datasets", version = "5.0.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "datasets", version = "5.1.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "decord2", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'arm64' and sys_platform == 'linux') or (platform_machine == 'armv7l' and sys_platform == 'linux')" }, { name = "easydict" }, { name = "einops" }, { name = "fastapi" }, { name = "flash-attn-4" }, - { name = "flashinfer-cubin", version = "0.6.11.post1", source = { registry = "https://pypi.org/simple" } }, + { name = "flashinfer-cubin" }, { name = "flashinfer-python", version = "0.6.11.post1", source = { registry = "https://pypi.org/simple" } }, { name = "gguf" }, { name = "interegular" }, @@ -8818,7 +10455,8 @@ dependencies = [ { name = "ipython", version = "9.12.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "kernels" }, { name = "llguidance", version = "0.7.30", source = { registry = "https://pypi.org/simple" } }, - { name = "mistral-common" }, + { name = "mistral-common", version = "1.11.2", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "mistral-common", version = "1.12.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "modelscope" }, { name = "msgspec" }, { name = "ninja" }, @@ -8826,7 +10464,7 @@ dependencies = [ { name = "nvidia-cutlass-dsl", version = "4.5.1", source = { registry = "https://pypi.org/simple" }, extra = ["cu13"], marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "nvidia-ml-py" }, { name = "openai", version = "2.6.1", source = { registry = "https://pypi.org/simple" } }, - { name = "openai-harmony", version = "0.0.4", source = { registry = "https://pypi.org/simple" } }, + { name = "openai-harmony" }, { name = "orjson" }, { name = "outlines" }, { name = "packaging" }, @@ -8840,7 +10478,7 @@ dependencies = [ { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "python-multipart" }, { name = "pyzmq" }, - { name = "quack-kernels" }, + { name = "quack-kernels", version = "0.4.1", source = { registry = "https://pypi.org/simple" } }, { name = "requests" }, { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, @@ -8854,12 +10492,12 @@ dependencies = [ { name = "tilelang", version = "0.1.8", source = { registry = "https://pypi.org/simple" } }, { name = "timm" }, { name = "tokenspeed-mla", version = "0.1.1", source = { registry = "https://pypi.org/simple" } }, - { name = "torch" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" } }, { name = "torch-memory-saver" }, { name = "torchao" }, { name = "torchaudio" }, { name = "torchcodec", version = "0.11.1", source = { registry = "https://pypi.org/simple" }, marker = "(platform_machine != 'aarch64' and platform_machine != 'arm64' and platform_machine != 'armv7l') or sys_platform != 'linux'" }, - { name = "torchvision" }, + { name = "torchvision", version = "0.26.0", source = { registry = "https://pypi.org/simple" } }, { name = "tqdm" }, { name = "transformers", version = "5.6.0", source = { registry = "https://pypi.org/simple" } }, { name = "uvicorn" }, @@ -8923,7 +10561,8 @@ version = "0.4.14" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "grpcio" }, - { name = "protobuf" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/a7/d4/beb65e474390131753f9cf984f6ad90e98af46d74836adce78eaf1291eb7/smg_grpc_proto-0.4.14.tar.gz", hash = "sha256:a8ea20872934472c94ffa192bafb2a4f71bc591ab4bfb240cb567801210fbc0f", size = 28677, upload-time = "2026-07-11T08:16:55.404Z" } wheels = [ @@ -8936,8 +10575,10 @@ version = "0.8.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "grpcio" }, - { name = "grpcio-health-checking" }, - { name = "grpcio-reflection" }, + { name = "grpcio-health-checking", version = "1.81.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "grpcio-health-checking", version = "1.83.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "grpcio-reflection", version = "1.81.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "grpcio-reflection", version = "1.83.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "smg-grpc-proto" }, ] sdist = { url = "https://files.pythonhosted.org/packages/7e/7d/617cb9261856613f491b5fd402543dc40cc371156cd020073552d7e9ea86/smg_grpc_servicer-0.8.0.tar.gz", hash = "sha256:d7e730d8b2f44ab8378b4293fde1d8fa1037a3bacdf56984c91912e17a336f52", size = 109736, upload-time = "2026-07-30T05:51:13.273Z" } @@ -8977,8 +10618,8 @@ name = "soundfile" version = "0.13.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "cffi" }, - { name = "numpy" }, + { name = "cffi", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, + { name = "numpy", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/e1/41/9b873a8c055582859b239be17902a85339bec6a30ad162f98c9b0288a2cc/soundfile-0.13.1.tar.gz", hash = "sha256:b2c68dab1e30297317080a5b43df57e302584c49e2942defdde0acccc53f0e5b", size = 46156, upload-time = "2025-01-25T09:17:04.831Z" } wheels = [ @@ -8996,7 +10637,7 @@ name = "soxr" version = "1.0.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy" }, + { name = "numpy", marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/42/7e/f4b461944662ad75036df65277d6130f9411002bfb79e9df7dff40a31db9/soxr-1.0.0.tar.gz", hash = "sha256:e07ee6c1d659bc6957034f4800c60cb8b98de798823e34d2a2bba1caa85a4509", size = 171415, upload-time = "2025-09-07T13:22:21.317Z" } wheels = [ @@ -9027,8 +10668,8 @@ name = "sse-starlette" version = "3.3.4" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "anyio" }, - { name = "starlette" }, + { name = "anyio", marker = "python_full_version >= '3.11'" }, + { name = "starlette", marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/26/8c/f9290339ef6d79badbc010f067cd769d6601ec11a57d78569c683fb4dd87/sse_starlette-3.3.4.tar.gz", hash = "sha256:aaf92fc067af8a5427192895ac028e947b484ac01edbc3caf00e7e7137c7bef1", size = 32427, upload-time = "2026-03-29T09:00:23.307Z" } wheels = [ @@ -9054,9 +10695,9 @@ name = "starlette" version = "1.3.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "anyio" }, + { name = "anyio", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.13' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and python_full_version < '3.13' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/eb/e3/7c1dc7381d9f8ab7d854328ebfa884e62cb3f3d8549ddfd37c7814f42afa/starlette-1.3.1.tar.gz", hash = "sha256:05d0213193f2fbaae60e2ecb593b4add4262ad4e46536b54abe36f11a71724e0", size = 2703240, upload-time = "2026-06-12T09:23:11.602Z" } wheels = [ @@ -9107,8 +10748,8 @@ name = "tiktoken" version = "0.12.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "regex" }, - { name = "requests" }, + { name = "regex", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, + { name = "requests", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/7d/ab/4d017d0f76ec3171d469d80fc03dfbb4e48a4bcaddaa831b31d526f05edc/tiktoken-0.12.0.tar.gz", hash = "sha256:b18ba7ee2b093863978fcb14f74b3707cdc8d4d4d3836853ce7ec60772139931", size = 37806, upload-time = "2025-10-06T20:22:45.419Z" } wheels = [ @@ -9186,7 +10827,7 @@ dependencies = [ { name = "numpy" }, { name = "psutil" }, { name = "setuptools", marker = "sys_platform == 'darwin'" }, - { name = "torch" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" } }, { name = "torch-c-dlpack-ext", marker = "python_full_version < '3.14'" }, { name = "tqdm" }, { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, @@ -9202,41 +10843,49 @@ wheels = [ [[package]] name = "tilelang" -version = "0.1.9" +version = "0.1.12" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32'", - "python_full_version >= '3.14' and sys_platform == 'emscripten'", - "python_full_version >= '3.14' and sys_platform == 'darwin'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", - "python_full_version == '3.13.*' and sys_platform == 'darwin'", - "python_full_version == '3.13.*' and sys_platform != 'darwin'", - "python_full_version == '3.12.*' and sys_platform == 'darwin'", - "python_full_version == '3.12.*' and sys_platform != 'darwin'", - "python_full_version == '3.11.*' and sys_platform == 'darwin'", - "python_full_version == '3.11.*' and sys_platform != 'darwin'", - "python_full_version < '3.11' and sys_platform == 'darwin'", - "python_full_version < '3.11' and sys_platform != 'darwin'", -] -dependencies = [ - { name = "apache-tvm-ffi", version = "0.1.9", source = { registry = "https://pypi.org/simple" } }, - { name = "cloudpickle" }, - { name = "ml-dtypes" }, - { name = "numpy" }, - { name = "psutil" }, - { name = "setuptools", marker = "sys_platform == 'darwin'" }, - { name = "torch" }, - { name = "torch-c-dlpack-ext", marker = "python_full_version < '3.14'" }, - { name = "tqdm" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "z3-solver" }, + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin'", +] +dependencies = [ + { name = "apache-tvm-ffi", version = "0.1.11", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "cloudpickle", marker = "python_full_version >= '3.11'" }, + { name = "ml-dtypes", marker = "python_full_version >= '3.11'" }, + { name = "numpy", marker = "python_full_version >= '3.11'" }, + { name = "psutil", marker = "python_full_version >= '3.11'" }, + { name = "setuptools", marker = "python_full_version >= '3.11' and sys_platform == 'darwin'" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "torch-c-dlpack-ext", marker = "python_full_version >= '3.11' and python_full_version < '3.14'" }, + { name = "tqdm", marker = "python_full_version >= '3.11'" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "z3-solver", marker = "python_full_version >= '3.11'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/56/70/5051f65821baa30a3d61fc48f8ba10c776490315e8c90f82559b92089756/tilelang-0.1.9.tar.gz", hash = "sha256:287f727c913bb648fcf6c1968809ba3390e55eeed257a5c6bb9a80bc05966af4", size = 93395292, upload-time = "2026-04-22T09:19:11.988Z" } +sdist = { url = "https://files.pythonhosted.org/packages/6d/9a/5776adb5b6e03e141c8cff020e16c45b87c9b7a7a5bdfe83416d6cc23933/tilelang-0.1.12.tar.gz", hash = "sha256:595b39581d099a53f875bd8d553863e7e4f58998052b58764f5472cbfbb87043", size = 93565090, upload-time = "2026-07-08T04:22:52.901Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/90/db/4dd76da8c8585c605639a21bc098d504e317fe324a72f01ce3c7370250b4/tilelang-0.1.9-cp38-abi3-macosx_11_0_arm64.whl", hash = "sha256:00ed594fdeb229c5505b9ffa895c3c5daeb28641c78f783fa1f724cf1e08cecd", size = 36599020, upload-time = "2026-04-22T09:14:39.366Z" }, - { url = "https://files.pythonhosted.org/packages/f7/8a/1cbeee79d62abaa02441c2d00621554e41aa62dbf3b94a4feb3867184b01/tilelang-0.1.9-cp38-abi3-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4bbccfe9035aed775ffafb6dc25a5994504b24e2c5d95d0f39643edfafa7bf12", size = 45419374, upload-time = "2026-04-22T09:15:56.014Z" }, - { url = "https://files.pythonhosted.org/packages/c6/a7/f4bfb86f87e107703146e703204cec2c0eae2492b633e0052b0ace3febb6/tilelang-0.1.9-cp38-abi3-manylinux_2_34_aarch64.whl", hash = "sha256:77ab0ee2f40f66ea015b6b21426d482751e28cbc635ef9d1198cbd6502454a7c", size = 42110365, upload-time = "2026-04-22T09:17:18.292Z" }, + { url = "https://files.pythonhosted.org/packages/13/51/4ae1fb532406082a4ebf16c890f5f98db14b8588f2c486334ba23578b4bc/tilelang-0.1.12-cp38-abi3-macosx_11_0_arm64.whl", hash = "sha256:8741d6cd519782645d55c17abd8986402c62948bc2122d3417c445d11de5dd16", size = 38347295, upload-time = "2026-07-08T04:22:35.691Z" }, + { url = "https://files.pythonhosted.org/packages/d1/53/f281a0bd9ee7e03d6a97828fc0e443321ed26ea2b0bd74bf9f1d9451d30f/tilelang-0.1.12-cp38-abi3-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:bbeb5573cbe2544a51a5c9a6cc737c5e3b5c2d95411caa3687fd9b08eb9d5f97", size = 50501074, upload-time = "2026-07-08T04:22:39.973Z" }, + { url = "https://files.pythonhosted.org/packages/fb/a4/ff0b14d00d29fe418063cd97490f55aa9cad6f981a6fbbf29124b024637d/tilelang-0.1.12-cp38-abi3-manylinux_2_34_aarch64.whl", hash = "sha256:a1701eba4ad603d73f1b128b88f6a2350f1b1161729e5b2d1e540a1ec7a82d52", size = 46281933, upload-time = "2026-07-08T04:22:44.244Z" }, + { url = "https://files.pythonhosted.org/packages/cb/15/1a8a5b5df0a23ece80b4da8e49c43418244e5ab165ed1938a1cc2656cda2/tilelang-0.1.12-cp38-abi3-win_amd64.whl", hash = "sha256:daeb3d529ab9e7665adb94c4bf0d47732294709aa9b8b9bd673fa48001385422", size = 34020025, upload-time = "2026-07-08T04:22:48.066Z" }, ] [[package]] @@ -9244,11 +10893,13 @@ name = "timm" version = "1.0.16" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "huggingface-hub" }, + { name = "huggingface-hub", version = "1.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "huggingface-hub", version = "1.33.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pyyaml" }, - { name = "safetensors" }, - { name = "torch" }, - { name = "torchvision" }, + { name = "safetensors", version = "0.7.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "safetensors", version = "0.8.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" } }, + { name = "torchvision", version = "0.26.0", source = { registry = "https://pypi.org/simple" } }, ] sdist = { url = "https://files.pythonhosted.org/packages/94/f6/4d7a8c261341fa6ad281920618739f2a650f41043afcedb570f24e99a776/timm-1.0.16.tar.gz", hash = "sha256:a3b8130dd2cb8dc3b9f5e3d09ab6d677a6315a8695fd5264eb6d52a4a46c1044", size = 2339999, upload-time = "2025-06-26T17:09:44.208Z" } wheels = [ @@ -9268,8 +10919,24 @@ wheels = [ name = "tokenizers" version = "0.22.2" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] dependencies = [ - { name = "huggingface-hub" }, + { name = "huggingface-hub", version = "1.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "huggingface-hub", version = "1.33.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/73/6f/f80cfef4a312e1fb34baf7d85c72d4411afde10978d4657f8cdd811d3ccc/tokenizers-0.22.2.tar.gz", hash = "sha256:473b83b915e547aa366d1eee11806deaf419e17be16310ac0a14077f1e28f917", size = 372115, upload-time = "2026-01-05T10:45:15.988Z" } wheels = [ @@ -9294,6 +10961,64 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/59/8c/b1c87148aa15e099243ec9f0cf9d0e970cc2234c3257d558c25a2c5304e6/tokenizers-0.22.2-pp310-pypy310_pp73-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f01a9c019878532f98927d2bacb79bbb404b43d3437455522a00a30718cdedb5", size = 3373542, upload-time = "2026-01-05T10:40:52.803Z" }, ] +[[package]] +name = "tokenizers" +version = "0.23.3" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "huggingface-hub", version = "2.3.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/e0/7c/2cabb2174e772636683008f2c5621949b645da7d303c596589e84516a184/tokenizers-0.23.3.tar.gz", hash = "sha256:cded33237c77caeef62944d32aa9a7ef42bdce2b3497e18d137e072a8c4be438", size = 385286, upload-time = "2026-10-09T10:16:55.759Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/aa/2e/4ce5b9716f26e526eff6b0502ebed4ea8d7161f03b3c77617c9f25528e97/tokenizers-0.23.3-cp310-abi3-macosx_10_12_x86_64.whl", hash = "sha256:9d2b5c97daf61688c2ad1803ca851800feaba50fb68d5821779e9ea5880d968c", size = 3148800, upload-time = "2026-10-09T10:00:51.457Z" }, + { url = "https://files.pythonhosted.org/packages/b2/72/01e49f032bb346e5aaf06c10c74fe8aeec847173adbadd66eb7c53054bf2/tokenizers-0.23.3-cp310-abi3-macosx_11_0_arm64.whl", hash = "sha256:68649e97d5b43c44c031d8d848874a6eecae8f8fe40ea989aa777a5a83aca716", size = 3101381, upload-time = "2026-10-09T10:00:54.063Z" }, + { url = "https://files.pythonhosted.org/packages/15/fc/ae987741829b1cd547668c4c94be732ae3eefd1d74344e64c3d2ca714acd/tokenizers-0.23.3-cp310-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ec82e80e65a862275b97c3d90b7a523df8d9519ee48aeb4e9625b2cc909274e0", size = 3519944, upload-time = "2026-10-09T10:00:55.885Z" }, + { url = "https://files.pythonhosted.org/packages/1c/da/cc8f6c030afaf05fbddc608158fbb761dca46913cbeba6b112e59fc82e2a/tokenizers-0.23.3-cp310-abi3-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:c64a0713180ff16829d4e7f39a658b77ea11443af4e1aa46523692943c9b1414", size = 3397695, upload-time = "2026-10-09T10:00:57.444Z" }, + { url = "https://files.pythonhosted.org/packages/ec/f1/256f78d1365fa2cd3ea6db716883d74667c8cbb6a21f15fa5b89a773cdc2/tokenizers-0.23.3-cp310-abi3-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ddedfd4b3b4be6be24ff6ca645c4a37fddfd305f6f3e354c54cf10b715c48215", size = 3753125, upload-time = "2026-10-09T10:01:00.165Z" }, + { url = "https://files.pythonhosted.org/packages/60/93/eee007ac2fcbf4ecfce7fbc354826cf3611f56bdb886f3e91b1f7dd06b8f/tokenizers-0.23.3-cp310-abi3-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:2a89614730d7b80940a5d2ed9320e1ec8add5a745c6151d8d05071b7215505b6", size = 4018598, upload-time = "2026-10-09T10:01:02.05Z" }, + { url = "https://files.pythonhosted.org/packages/bf/f9/0c96c4739461fce9d8d865b416728081bf6230022d7163bd6244f35f4b31/tokenizers-0.23.3-cp310-abi3-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:e88646b8580c5ad7f4361477f1298e9cc01771a1ee9aecfe32c47b8ff614cc38", size = 3602442, upload-time = "2026-10-09T10:01:03.77Z" }, + { url = "https://files.pythonhosted.org/packages/3a/40/6706b82693715581457c6d5423eaa7faae576bb0526c5738a57085eb4449/tokenizers-0.23.3-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:376851d22bcf9d650a5c3090bb83e6cf9e895fbf0595369fa4cd43c1f69b5f87", size = 3396193, upload-time = "2026-10-09T10:01:05.48Z" }, + { url = "https://files.pythonhosted.org/packages/fe/0c/85946de40e25b7364b8f1bcf56def129069acd5bb364b7c86a32919e1a23/tokenizers-0.23.3-cp310-abi3-manylinux_2_31_riscv64.whl", hash = "sha256:bf501c40b72d2d5c8623620210430e9cac1ce47a46e45b34107b70a1557d46b0", size = 3553483, upload-time = "2026-10-09T10:01:07.387Z" }, + { url = "https://files.pythonhosted.org/packages/f1/6b/8d615d92cad1d511ca5ab188d1c7c167f0b3d295cc0d96207f9f82d486d8/tokenizers-0.23.3-cp310-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:114e2b55ed177179d59f4ab98200a4471e11e78f9e4b5a922d146740f96fcf52", size = 9972248, upload-time = "2026-10-09T10:01:09.437Z" }, + { url = "https://files.pythonhosted.org/packages/c9/7d/a922e37ddd58d1b463bbc2ad08120c8f59c60b814cd353519a116b24f8ba/tokenizers-0.23.3-cp310-abi3-musllinux_1_2_armv7l.whl", hash = "sha256:d3407fb7b9c4d75dd68850ffd7180bc0a5d2dbaf0762d888e612f31fec3f9c6b", size = 9802957, upload-time = "2026-10-09T10:01:11.869Z" }, + { url = "https://files.pythonhosted.org/packages/4b/06/5d3f506a86ae0699a0e4ea05c05978f9aee169ef2c1d844e68c971cf8194/tokenizers-0.23.3-cp310-abi3-musllinux_1_2_i686.whl", hash = "sha256:84513ef0aeb8bf8f4ea11a2e8a7ac163ec5288aa115e649a59b470ac5c3107df", size = 10145487, upload-time = "2026-10-09T10:01:14.268Z" }, + { url = "https://files.pythonhosted.org/packages/26/e5/065625317690ea3548d834dad81f48ea1fd32e4964610e658e195d7fe28e/tokenizers-0.23.3-cp310-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:e05ab7baf7f47b406a95fea6f3b0a484b2ddcd9e1d14b68844c457eb755085a3", size = 10266026, upload-time = "2026-10-09T10:16:33.054Z" }, + { url = "https://files.pythonhosted.org/packages/77/4e/babede85d0d19f5e3deeef0063e01848141329934d3d77c31b5cab5ac2b4/tokenizers-0.23.3-cp310-abi3-win32.whl", hash = "sha256:1ebf28794e7e4954e20a7f70fbea410b2d1f0418f7dbbca97ca384fcfef38c25", size = 2588086, upload-time = "2026-10-09T10:16:35.686Z" }, + { url = "https://files.pythonhosted.org/packages/d1/6c/24f074c9a0efb98e61b20aafe6b2641922d5db24e447d5d6daffd9e17555/tokenizers-0.23.3-cp310-abi3-win_amd64.whl", hash = "sha256:1f0823bb00c5fdc98e487354d54dd55a03848d61a1a0bf29a68c77f24f3b26c3", size = 2872101, upload-time = "2026-10-09T10:16:37.533Z" }, + { url = "https://files.pythonhosted.org/packages/53/77/a476b6f73a661c11d113a342d2326b91506cf2285f0995d1212a6bb2022d/tokenizers-0.23.3-cp310-abi3-win_arm64.whl", hash = "sha256:7e48734d2de9260d86f03ab056d2cfeeff3869f61dbd49aaa15a2793b5f3458b", size = 2742580, upload-time = "2026-10-09T10:16:39.244Z" }, + { url = "https://files.pythonhosted.org/packages/65/46/f66baaedd42414a3f583c47379dc350e3e1f858a690d2574fd85ae70681b/tokenizers-0.23.3-cp314-cp314t-macosx_10_12_x86_64.whl", hash = "sha256:efa3d7318406b4d115dce61ad5061953f1f44b128e79c020ce4615d763e23b6e", size = 3154274, upload-time = "2026-10-09T10:16:40.876Z" }, + { url = "https://files.pythonhosted.org/packages/c6/41/8de8c63b2d935eee5a0f42011fb7b786ffafeab0b8eb6d17acb8af2293b7/tokenizers-0.23.3-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:a4fbb3662f9f59d199d61338e54b4bcc11d07ebbb1aeb3540dacb2be9c521cb7", size = 3077805, upload-time = "2026-10-09T10:16:42.856Z" }, + { url = "https://files.pythonhosted.org/packages/e3/08/b1cbae8dc8fc7c91f992ac2d87a086e9b3f25a28814047ca16a82fe8c87b/tokenizers-0.23.3-cp314-cp314t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:de536665495cb4b409d25bade41963f801aff4225c19a6b804b048f7d14e34c7", size = 3491678, upload-time = "2026-10-09T10:16:45.093Z" }, + { url = "https://files.pythonhosted.org/packages/3e/0d/aac0cb2f3a1fdbef514145b4c5f2df4d05deeb1ee8f73ae641a1b4a62a85/tokenizers-0.23.3-cp314-cp314t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:5cc24bb457dd4a8af89c8fcb40074d570129ec473df2a866c276ee55db4749d7", size = 3367420, upload-time = "2026-10-09T10:16:47.112Z" }, + { url = "https://files.pythonhosted.org/packages/1e/1d/41a697d0c193a320b243fbd68b2057b6eb2f01ecf80899e1a16e646ff699/tokenizers-0.23.3-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:acd5c57b4bd3e56e246e2731a3a3a6825a7a7d89b7e3b761ba80bc521710f04b", size = 9945973, upload-time = "2026-10-09T10:16:49.326Z" }, + { url = "https://files.pythonhosted.org/packages/37/e9/b56e619fcd583000a2b1254bb46af8dc6a174d3ba3329f454ad5a95a2be2/tokenizers-0.23.3-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:82eb480f6f1c21cea3349dec32cf1a6384c6c1e775f00f83b0d51197bc013687", size = 10237491, upload-time = "2026-10-09T10:16:51.943Z" }, + { url = "https://files.pythonhosted.org/packages/6f/68/f58b3beb95f3b62816e91e5e768e684cd63e58f9cbece22036dae3b1c971/tokenizers-0.23.3-cp314-cp314t-win_amd64.whl", hash = "sha256:1554a6eed34d9d6a78d23360f4e06df8dffab1ae08c7e8488e0b3e3b36cc266f", size = 2847654, upload-time = "2026-10-09T10:16:54.166Z" }, +] + [[package]] name = "tokenspeed-mla" version = "0.1.1" @@ -9313,8 +11038,9 @@ resolution-markers = [ dependencies = [ { name = "apache-tvm-ffi", version = "0.1.9", source = { registry = "https://pypi.org/simple" } }, { name = "nvidia-cutlass-dsl", version = "4.5.1", source = { registry = "https://pypi.org/simple" } }, - { name = "tokenspeed-triton" }, - { name = "torch" }, + { name = "tokenspeed-triton", version = "3.7.10.post20260505", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tokenspeed-triton", version = "3.9.10.post20261008", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" } }, ] sdist = { url = "https://files.pythonhosted.org/packages/bf/71/659b972967ce37af1795b3bd83c6605d0dd3e1e49986bb227aca3925a82e/tokenspeed_mla-0.1.1.tar.gz", hash = "sha256:df67032b47460860676fca3592f5637c2cf70baf708e32150022e23493dc6883", size = 119154, upload-time = "2026-05-05T21:44:03.874Z" } wheels = [ @@ -9323,37 +11049,40 @@ wheels = [ [[package]] name = "tokenspeed-mla" -version = "0.1.2" +version = "0.2.15" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32'", - "python_full_version >= '3.14' and sys_platform == 'emscripten'", - "python_full_version >= '3.14' and sys_platform == 'darwin'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", - "python_full_version == '3.13.*' and sys_platform == 'darwin'", - "python_full_version == '3.13.*' and sys_platform != 'darwin'", - "python_full_version == '3.12.*' and sys_platform == 'darwin'", - "python_full_version == '3.12.*' and sys_platform != 'darwin'", - "python_full_version == '3.11.*' and sys_platform == 'darwin'", - "python_full_version == '3.11.*' and sys_platform != 'darwin'", - "python_full_version < '3.11' and sys_platform == 'darwin'", - "python_full_version < '3.11' and sys_platform != 'darwin'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin'", ] dependencies = [ - { name = "apache-tvm-ffi", version = "0.1.9", source = { registry = "https://pypi.org/simple" } }, - { name = "nvidia-cutlass-dsl", version = "4.5.2", source = { registry = "https://pypi.org/simple" } }, - { name = "tokenspeed-triton" }, - { name = "torch" }, + { name = "apache-tvm-ffi", version = "0.1.11", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version >= '3.11' and python_full_version < '3.14' and sys_platform == 'win32') or (python_full_version >= '3.11' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32')" }, + { name = "nvidia-cutlass-dsl", version = "4.8.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version >= '3.11' and python_full_version < '3.14' and sys_platform == 'win32') or (python_full_version >= '3.11' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32')" }, + { name = "tokenspeed-triton", version = "3.9.10.post20261008", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version >= '3.11' and python_full_version < '3.14' and sys_platform == 'win32') or (python_full_version >= '3.11' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32')" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version >= '3.11' and python_full_version < '3.14' and sys_platform == 'win32') or (python_full_version >= '3.11' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32')" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/42/20/4110d624d81d63f0bee2f19dba7ea0e1d8a31ea50147e6c1db82223c88a4/tokenspeed_mla-0.1.2-py3-none-manylinux_2_28_aarch64.whl", hash = "sha256:592590f36d85e624ecdc5e357ff35e29e761e6d879900dce8b67a6785c8ce75c", size = 743769, upload-time = "2026-05-13T03:30:54.486Z" }, - { url = "https://files.pythonhosted.org/packages/84/01/4bf8b74ead3e8e7c1c809435396254c067a33fde48acc20f602aae622d97/tokenspeed_mla-0.1.2-py3-none-manylinux_2_28_x86_64.whl", hash = "sha256:c9466a351fe039792e56cf49f3e79744c1dc28c7af10306a02e62b8e92fa5985", size = 748681, upload-time = "2026-05-13T03:30:56.718Z" }, + { url = "https://files.pythonhosted.org/packages/14/56/495ed0feea7c332e02bbf23fa565aee3a176fe463c81ff7cc85afc01817e/tokenspeed_mla-0.2.15-py3-none-any.whl", hash = "sha256:7b4d8bfe92e1ad984e94db262671ef33ce2ecfe0a44f79c5ea1c2816d8397a1f", size = 151477, upload-time = "2026-09-28T19:49:06.103Z" }, ] [[package]] name = "tokenspeed-triton" version = "3.7.10.post20260505" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform != 'darwin'", +] wheels = [ { url = "https://files.pythonhosted.org/packages/80/fc/7fc1fa2be130cd61b1422753b3b8a526c8e59b9ad1e78a0af34872ab7d82/tokenspeed_triton-3.7.10.post20260505-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:82c222755095db261e32e3964e009573f3360806088fa493be65404276866344", size = 82061800, upload-time = "2026-05-05T07:48:43.527Z" }, { url = "https://files.pythonhosted.org/packages/47/c6/7640856932343ab4bebf7eddbbd8b7537a978a1b525ed70d1ed76e01b293/tokenspeed_triton-3.7.10.post20260505-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:060f657c78b5cd0c5645f01eb0f73b72cf385589235e3b96ca05f9b3d33a644f", size = 89124672, upload-time = "2026-05-05T07:48:47.896Z" }, @@ -9363,6 +11092,31 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/af/c3/4808d86016368fed9495c3a3408cc7f912e7863ff3432937404bd0a551a6/tokenspeed_triton-3.7.10.post20260505-cp312-abi3-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:19618c7db01a9bd33885f7acbf8945adb2f5534668aa97629b56d481753cbcad", size = 89127692, upload-time = "2026-05-05T07:49:04.22Z" }, ] +[[package]] +name = "tokenspeed-triton" +version = "3.9.10.post20261008" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/b3/d8/beb184482fc9370af7ae8fd169911c87e3e783f202b6393b87606d568cfd/tokenspeed_triton-3.9.10.post20261008-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:3044af3d696144a6c48e13519c136a3427d0db11818dc45dbc17c79bdc35c35f", size = 88575808, upload-time = "2026-10-09T23:22:57.279Z" }, + { url = "https://files.pythonhosted.org/packages/a8/fd/d32c81856f49ff8708019898632a785f583c533b84350425bb7fda57fc61/tokenspeed_triton-3.9.10.post20261008-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:96dbd5b519b22c9d3b4e39fa008f0c0b3b952f9b604a87578705c078d37af7b2", size = 93282180, upload-time = "2026-10-09T23:23:01.496Z" }, + { url = "https://files.pythonhosted.org/packages/b4/e3/b5a3fc7df1c9996c4423deef5e3bad912c27be09922ad9fcb902ab960fc9/tokenspeed_triton-3.9.10.post20261008-cp312-abi3-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:e6bd2c4b25f6976f1e22e33e5e1c8b04ede6db6aa6c50722b8a30097ec2782ad", size = 88581214, upload-time = "2026-10-09T23:23:05.448Z" }, + { url = "https://files.pythonhosted.org/packages/1c/79/ffd666bee268dc27ee34ed471e43f0357f9154c93d5180381861f8673c21/tokenspeed_triton-3.9.10.post20261008-cp312-abi3-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5cb864b8a55f3b0b6d95dca81af85abd702496265971d8c44366eb68c8851154", size = 93285278, upload-time = "2026-10-09T23:23:09.368Z" }, +] + [[package]] name = "tomli" version = "2.4.1" @@ -9430,23 +11184,38 @@ wheels = [ name = "torch" version = "2.11.0" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] dependencies = [ - { name = "cuda-bindings", marker = "sys_platform == 'linux'" }, - { name = "cuda-toolkit", extra = ["cublas", "cudart", "cufft", "cufile", "cupti", "curand", "cusolver", "cusparse", "nvjitlink", "nvrtc", "nvtx"], marker = "sys_platform == 'linux'" }, + { name = "cuda-bindings", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cuda-toolkit", version = "13.0.2", source = { registry = "https://pypi.org/simple" }, extra = ["cublas", "cudart", "cufft", "cufile", "cupti", "curand", "cusolver", "cusparse", "nvjitlink", "nvrtc", "nvtx"], marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "filelock", version = "4.1.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "filelock", version = "4.2.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, - { name = "fsspec" }, - { name = "jinja2" }, + { name = "filelock", version = "4.2.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "fsspec", marker = "python_full_version < '3.11' or extra == 'extra-7-vec-inf-sglang'" }, + { name = "jinja2", marker = "python_full_version < '3.11' or extra == 'extra-7-vec-inf-sglang'" }, { name = "networkx", version = "3.4.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "networkx", version = "3.6.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, - { name = "nvidia-cudnn-cu13", marker = "sys_platform == 'linux'" }, - { name = "nvidia-cusparselt-cu13", marker = "sys_platform == 'linux'" }, - { name = "nvidia-nccl-cu13", marker = "sys_platform == 'linux'" }, - { name = "nvidia-nvshmem-cu13", marker = "sys_platform == 'linux'" }, - { name = "setuptools" }, - { name = "sympy" }, - { name = "triton", marker = "sys_platform == 'linux'" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "networkx", version = "3.6.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cudnn-cu13", version = "9.19.0.56", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cusparselt-cu13", version = "0.8.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-nccl-cu13", version = "2.28.9", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-nvshmem-cu13", marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "setuptools", marker = "python_full_version < '3.11' or extra == 'extra-7-vec-inf-sglang'" }, + { name = "sympy", marker = "python_full_version < '3.11' or extra == 'extra-7-vec-inf-sglang'" }, + { name = "triton", version = "3.6.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and sys_platform == 'linux') or (sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, ] wheels = [ @@ -9480,12 +11249,84 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/cf/bf/c8d12a2c86dbfd7f40fb2f56fbf5a505ccf2d9ce131eb559dfc7c51e1a04/torch-2.11.0-cp314-cp314t-win_amd64.whl", hash = "sha256:b2a43985ff5ef6ddd923bbcf99943e5f58059805787c5c9a2622bf05ca2965b0", size = 114792991, upload-time = "2026-03-23T18:08:19.216Z" }, ] +[[package]] +name = "torch" +version = "2.13.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "cuda-bindings", marker = "(python_full_version >= '3.11' and python_full_version < '3.15' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cuda-toolkit", version = "13.0.3.0", source = { registry = "https://pypi.org/simple" }, extra = ["cublas", "cudart", "cufft", "cufile", "cupti", "curand", "cusolver", "cusparse", "nvjitlink", "nvrtc", "nvtx"], marker = "(python_full_version >= '3.11' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "filelock", version = "4.2.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "fsspec", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "jinja2", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "networkx", version = "3.6.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cudnn-cu13", version = "9.20.0.48", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-cusparselt-cu13", version = "0.8.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-nccl-cu13", version = "2.29.7", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvidia-nvshmem-cu13", marker = "(python_full_version >= '3.11' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "setuptools", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "sympy", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "triton", version = "3.7.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and python_full_version < '3.15' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/7f/e7/19894fdb51c7dbaf94f5a79bb0871da0992e8e4241e579cb006da46d2e58/torch-2.13.0-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:94f0de129916f77b8dc2c7a8eff644cfeddfe59e39c9f55e9f6e17543410281d", size = 111178962, upload-time = "2026-07-08T16:05:49.855Z" }, + { url = "https://files.pythonhosted.org/packages/d1/5c/b1d5de470c54e339b30a92d96683a71bcebd78f5f2a7fc714cd6dc6bbd68/torch-2.13.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:0ab4b69f3ee03a62a002cfbf77b1ca5e88aceb4ea64cb4388bb28f638ddbb045", size = 427198333, upload-time = "2026-07-08T16:05:36.847Z" }, + { url = "https://files.pythonhosted.org/packages/50/c0/68a84105e1fcb8970144b388ff3d3e5dc15a3be28c1e247841f7d7247e41/torch-2.13.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:c78b7b4d04461855a764cf01bae9a462bb88bc93defcfa11235cbc8fdf3e12c4", size = 526555154, upload-time = "2026-07-08T16:05:06.507Z" }, + { url = "https://files.pythonhosted.org/packages/2e/c9/0bb9d097b03cbaf96bb75b15e867347b8e41bfcdfe0539452d17d9e63993/torch-2.13.0-cp310-cp310-win_amd64.whl", hash = "sha256:2bd30b6b730d987fa386ce3898933762c5cb8cc82eb0535211d787cc3ce2dfeb", size = 122015602, upload-time = "2026-07-08T16:05:45.25Z" }, + { url = "https://files.pythonhosted.org/packages/5b/fe/cba54dc58523434919b66f13a667e36e436deddd77ca519e96553617d4ec/torch-2.13.0-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:e76f9bcecc52b8ff711239a2f7547d5353df95878ab232f0773c1d95928b92f8", size = 111187938, upload-time = "2026-07-08T16:05:17.065Z" }, + { url = "https://files.pythonhosted.org/packages/c2/59/1e3160e18e12aa3038390efab3ce02b36a9d4d6a527ecdd8520dca2e68d8/torch-2.13.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:092790c696a760c729fd5722835f50b9d81fd7c8f141571f3f3cf4081a8f664c", size = 427199369, upload-time = "2026-07-08T16:04:51.054Z" }, + { url = "https://files.pythonhosted.org/packages/01/79/1f2d34ad7034ee1c7ffc1cf8bf0f8213af2a81df6ecdb3997ecec107c09d/torch-2.13.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:60fcdcb2f3876e21146cb4524ef06397d727ca9ad5f020818547e25075fe3cb7", size = 526574961, upload-time = "2026-07-08T16:04:07.075Z" }, + { url = "https://files.pythonhosted.org/packages/6c/fd/0f2ce40f58aefbdb3392f9acce3c8171940943ae2d661f70558bfa73befb/torch-2.13.0-cp311-cp311-win_amd64.whl", hash = "sha256:a0d8b11f16a48d60e2015d8213aa0390744cbebb98e58b62b3514dddc656e330", size = 122015870, upload-time = "2026-07-08T16:05:27.59Z" }, + { url = "https://files.pythonhosted.org/packages/c4/3a/ed0f4d4d1dcde03bced7aac9a28e800abcdc0cbd06b6775044c9fbd877b7/torch-2.13.0-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:2fe228aba290d14b9f31b049be550dbd469c3fd3013d7a19705b30454da97027", size = 111213045, upload-time = "2026-07-08T16:05:22.997Z" }, + { url = "https://files.pythonhosted.org/packages/df/a9/f6a2a4d763ff1df02e9a64c477029db614295bc9367f4131223791ccc243/torch-2.13.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:572df8be8ffb4599c88cbd6a0726f1f854f4da65d2e3c09f0e2c2283333cd6d4", size = 427210998, upload-time = "2026-07-08T16:04:37.708Z" }, + { url = "https://files.pythonhosted.org/packages/f3/82/fea946351658e6534db52d2cc12bc53087cbf87f9440c5f180f367c1950b/torch-2.13.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:796633c4cdf0fe2cdced72d8f88f22e73dbcfce83132763162f6d4bff13b820b", size = 526605292, upload-time = "2026-07-08T16:04:22.81Z" }, + { url = "https://files.pythonhosted.org/packages/21/d6/e8f3c6f7e01f626f77259de9860d2a78bc84c40539e28e79b7e98b0bb659/torch-2.13.0-cp312-cp312-win_amd64.whl", hash = "sha256:024c6cc0c1b085f2f91f20a3dc27b0471d021c31ce84b81be3afdc39f791fd9d", size = 122057313, upload-time = "2026-07-08T16:03:53.43Z" }, + { url = "https://files.pythonhosted.org/packages/0d/fa/c1c10b7aff4a9a3e8956d4f0a5f468fa6db7abc3208805719076772b4833/torch-2.13.0-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:33449899ce5496c1b84b4853179d94fd102028ae1407314d9fb956bb79e70d09", size = 111213743, upload-time = "2026-07-08T16:03:28.579Z" }, + { url = "https://files.pythonhosted.org/packages/11/18/9ecb37b56293a0be8d80f810bf672a72fe7e02f8b475d5ef1b9bf8a0d748/torch-2.13.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:1e09d6a722504957c694faceca843acde562786df1144ebcc5a74075ec7f6005", size = 427213008, upload-time = "2026-07-08T16:03:44.106Z" }, + { url = "https://files.pythonhosted.org/packages/d4/5a/7c50ba1b7b713d71d34669c6d13dab0a11531a3eceb0307a5162dbfec0f7/torch-2.13.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:a3a9a21312872af8a26950b2c15680335a386a1f56ed03e780653d78b9607e9e", size = 526602329, upload-time = "2026-07-08T16:03:12.649Z" }, + { url = "https://files.pythonhosted.org/packages/91/3d/e7adcc6aaf36961cd18f56cf8ad0f3058c3a5c84ccf391762176c94581b8/torch-2.13.0-cp313-cp313-win_amd64.whl", hash = "sha256:49b58f1e2c52440abb6f17c28f0335fe6c6d01ad1a7f55b0183b81e4b34d64e6", size = 122057920, upload-time = "2026-07-08T16:03:01.808Z" }, + { url = "https://files.pythonhosted.org/packages/36/76/6dcc7f0c07052102dd36f83cbc5800842a909c8c3fbf1a7f8a5844954de9/torch-2.13.0-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:d849b390e07d8d333ce8ecaf91b273c656c598379a19c9acf1318a883f6b391c", size = 111227066, upload-time = "2026-07-08T16:03:33.6Z" }, + { url = "https://files.pythonhosted.org/packages/e9/09/2c10e8cd0e00fa5d23c052df6ce467eaa7182399f5e0f824f1e4ff42ccae/torch-2.13.0-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:a3893dc2da0a972a8ca5d698c85a9f967559ac5f8ee1797b77408aa8734d073c", size = 427226309, upload-time = "2026-07-08T16:02:53.127Z" }, + { url = "https://files.pythonhosted.org/packages/76/c6/22c2102bbef14ca6a6cb4c20e42f088e49c5f812be4e160ae57502e325f9/torch-2.13.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:49f1ea385c754e54919408a9bb3b5a72b0b755bbe2c916c1d6f70afbec4908a2", size = 526614507, upload-time = "2026-07-08T16:02:16.441Z" }, + { url = "https://files.pythonhosted.org/packages/2b/0c/7d1deb6bce5bc3e6042caf39100ac768eba3b9a098e1dddd16f75bd6489b/torch-2.13.0-cp314-cp314-win_amd64.whl", hash = "sha256:4f8573e3ce9ebcd53fe922f01077a6085ccdfbe5f12fd215883a9d87d7a744fd", size = 122051871, upload-time = "2026-07-08T16:03:23.521Z" }, + { url = "https://files.pythonhosted.org/packages/f4/ce/aa8b7f9949d32e0f2f624f342bc3b48112c1b8a130288465938bc83bcbf9/torch-2.13.0-cp314-cp314t-macosx_14_0_arm64.whl", hash = "sha256:c28def70706c2f9ecc752574766e8ae4da9b810ab6676b611166761a78a9f1e1", size = 111537025, upload-time = "2026-07-08T16:02:44.28Z" }, + { url = "https://files.pythonhosted.org/packages/69/d1/491e3a0389430946145888b0203f2b6a759ce2a61481b96a85c2da4f2ced/torch-2.13.0-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:31061ff56ed8fbf26c749806905aeb749ebeb819810fd5d52508aa5afd90dddc", size = 427219769, upload-time = "2026-07-08T16:02:31.18Z" }, + { url = "https://files.pythonhosted.org/packages/9a/1d/38006e045bf0a1fc28ef01e757c554e59e59a8770c284bc4f47b14e60441/torch-2.13.0-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:cc26eead4cf51d0b544e31e364dcf000846549c273bd148936fe9d24d29acb92", size = 526571320, upload-time = "2026-07-08T16:01:59.348Z" }, + { url = "https://files.pythonhosted.org/packages/56/94/655c91992a882bd5071aa0b5d22a07dbb130d801e872be97c0b627a7c693/torch-2.13.0-cp314-cp314t-win_amd64.whl", hash = "sha256:a7de8a313090dc5c7d7ba4bfe5c3be222528f9a4dba1acc83bddb1157360c4b8", size = 122306773, upload-time = "2026-07-08T16:02:39.832Z" }, +] + [[package]] name = "torch-c-dlpack-ext" version = "0.1.5" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "torch" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/37/de/921b6491efce5c389a5ef9bbed3d2d6660005840dae488124173180859ab/torch_c_dlpack_ext-0.1.5.tar.gz", hash = "sha256:d06f0357d575d22a168cc77acb9020fc4bae30968ceb6718a055dcbe92bacabe", size = 12913, upload-time = "2026-01-12T11:25:08.484Z" } wheels = [ @@ -9565,42 +11406,6 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/93/f7/ee5da8c03f1a3c7662c6c6a119f24a4b3e646da94be56dce3201e3a6ee9b/torchaudio-2.11.0-cp314-cp314t-win_amd64.whl", hash = "sha256:88fb5e29f670a33d9bac6aabb1d2734460cf6e461bde5cdc352826035851b16d", size = 328661, upload-time = "2026-03-23T18:13:20.1Z" }, ] -[[package]] -name = "torchcodec" -version = "0.9.1" -source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32'", - "python_full_version >= '3.14' and sys_platform == 'emscripten'", - "python_full_version >= '3.14' and sys_platform == 'darwin'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", - "python_full_version == '3.13.*' and sys_platform == 'darwin'", - "python_full_version == '3.13.*' and sys_platform != 'darwin'", - "python_full_version == '3.12.*' and sys_platform == 'darwin'", - "python_full_version == '3.12.*' and sys_platform != 'darwin'", - "python_full_version == '3.11.*' and sys_platform == 'darwin'", - "python_full_version == '3.11.*' and sys_platform != 'darwin'", - "python_full_version < '3.11' and sys_platform == 'darwin'", - "python_full_version < '3.11' and sys_platform != 'darwin'", -] -wheels = [ - { url = "https://files.pythonhosted.org/packages/4a/c3/ae904a2a8cb4b6288684f022a6d5061a74435bc2d6f1c4742c16d441a9ee/torchcodec-0.9.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:72738d7870edbcd6982207609efe314352e0bf90f4e3895a549e23a49ee0673c", size = 3397284, upload-time = "2025-12-10T15:55:45.498Z" }, - { url = "https://files.pythonhosted.org/packages/30/96/c693d75e32b4f07828c2b8a4b894be7366d0e46366ff10dad5825d891148/torchcodec-0.9.1-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:53dfecabc843e4d78b77453732129bd8572dd9d022ab81a94f5305cd5e920ded", size = 2043848, upload-time = "2025-12-10T15:55:25.772Z" }, - { url = "https://files.pythonhosted.org/packages/68/dd/cbda16eb33d380c3e7b3cb6dbcfa59aa21e676ddebe3627e834be3b1c124/torchcodec-0.9.1-cp310-cp310-win_amd64.whl", hash = "sha256:d5267e6112b0d0b948d653dc3ffa486bb9a01febc81606d3dfd75cf5a4dcc9f3", size = 2170387, upload-time = "2025-12-10T15:56:09.824Z" }, - { url = "https://files.pythonhosted.org/packages/c0/37/169238bb55017b08a93530d6f4474a112780df4c05c26c78df7dd58401f1/torchcodec-0.9.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:7a81d29e67cebefaec08f9c817635d1ff47205814884e527ef4b135a6c795612", size = 3899002, upload-time = "2025-12-10T15:55:49.911Z" }, - { url = "https://files.pythonhosted.org/packages/75/74/11c59e0592e555df78cfee2fd4b8ba5c725e4bf160002af88ce09de99d9d/torchcodec-0.9.1-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:5e959c7abf95de62f78653d416f00f7ca32936fbfd23371b23d5c8dc199f3670", size = 2048262, upload-time = "2025-12-10T15:55:29.552Z" }, - { url = "https://files.pythonhosted.org/packages/53/66/0612c7852cb7854dec45db9357397b4dbd4d6274832cc8949a8d1c0e518a/torchcodec-0.9.1-cp311-cp311-win_amd64.whl", hash = "sha256:508e0ce6da1e4af186b131cdbdd3d998bc4313e9f3a3f5b78f64aa5f9685ed88", size = 2181959, upload-time = "2025-12-10T15:56:13.191Z" }, - { url = "https://files.pythonhosted.org/packages/4b/60/3bfa459e09987af08e188811b191437c9d8215a74f4d418be6ff7df87b5c/torchcodec-0.9.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:8996ec62b72c69545c30246df64df386d06d7ec7de0689be5d20dfc06aad6442", size = 4064264, upload-time = "2025-12-10T15:55:56.313Z" }, - { url = "https://files.pythonhosted.org/packages/17/c8/bfb74babec98aff11ab4f239b0901f39e1a93338b3438e842d864dc46935/torchcodec-0.9.1-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:a50568ce73b70395d113833fb07394c223f5546ef5d4fafe0fdcd91627fca270", size = 2061978, upload-time = "2025-12-10T15:55:33.415Z" }, - { url = "https://files.pythonhosted.org/packages/b2/12/c0bbf01b0ed52b69aaeed4af1043dc8308ccc522a47fcc082b34882e2ba2/torchcodec-0.9.1-cp312-cp312-win_amd64.whl", hash = "sha256:d9c8efe5845bde45a428f96493b4a041511f47f5bd53b333a0ad90426be4623a", size = 2187178, upload-time = "2025-12-10T15:56:16.968Z" }, - { url = "https://files.pythonhosted.org/packages/6f/c7/67fc8417f9efa8a25c00a44f0d674761a0bad9c45e9725e3fd116b3c48ed/torchcodec-0.9.1-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:5c9cdcba50c75be70ef6ec919ec1f7f14d9d5163d93cf6bd94403e134f03734c", size = 4034415, upload-time = "2025-12-10T15:56:02.04Z" }, - { url = "https://files.pythonhosted.org/packages/68/05/06240f661e9aa08b20765305e3b88f60bff706bbe54ac35830af74612443/torchcodec-0.9.1-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:0643c5e9c3a51fdafdea87935d5b0a38e99626c664f47a150482d77ab370a877", size = 2067767, upload-time = "2025-12-10T15:55:37.27Z" }, - { url = "https://files.pythonhosted.org/packages/13/a2/d78cd65863fb805d9e35fe90ae7574eab86ff0ae63438208bd07d2cf1fd2/torchcodec-0.9.1-cp313-cp313-win_amd64.whl", hash = "sha256:df0b5a15998fd7457625c2af2a6276e0e710fac158d145045340dbbcd1cfdb65", size = 2186788, upload-time = "2025-12-10T15:56:20.204Z" }, - { url = "https://files.pythonhosted.org/packages/01/02/f8ae9443d3bcbe8a8d6d0bbc3992296e5476e5afa1f244100a3a7967a36c/torchcodec-0.9.1-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:b9bc5a5dff925df96d11bf90bd0ce964b8086bb11ae09adf353518192b5da483", size = 3812248, upload-time = "2025-12-10T15:56:06.382Z" }, - { url = "https://files.pythonhosted.org/packages/59/a1/8462b55571286847ea31edb7634583125400824267db9ba8301f4ce3f137/torchcodec-0.9.1-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:65634bb28b3155cf99f980dac31ecedb414c07b8156f8473ec9fb74bedbd2a1f", size = 2068456, upload-time = "2025-12-10T15:55:40.577Z" }, - { url = "https://files.pythonhosted.org/packages/f2/63/752d0fc1c6e8f799ae880ca1087510def663a7f9aa1a70074ae334c6908f/torchcodec-0.9.1-cp314-cp314-win_amd64.whl", hash = "sha256:2d01c8b3685a3a38f050ed2b526808a2938dba6f56cb9f9e967884fd858bba15", size = 2188320, upload-time = "2025-12-10T15:56:24.63Z" }, -] - [[package]] name = "torchcodec" version = "0.11.1" @@ -9640,14 +11445,79 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/18/96/6ee0e26547976dc55a69042ce895747a34221eab348931e975141d80d25e/torchcodec-0.11.1-cp314-cp314-win_amd64.whl", hash = "sha256:3fd9ef8302b261d3db5585e42be4a3138c5c240a822031642cdf1f82ea3db5b7", size = 1925002, upload-time = "2026-04-14T18:25:11.718Z" }, ] +[[package]] +name = "torchcodec" +version = "0.17.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version < '3.11' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform != 'darwin'", +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/73/8a/b44bbe3b9891dc8025518aec5837a31f735c9db288680acf9e95ea0276c9/torchcodec-0.17.0-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:645ea2b33596d3f1463a8a70a907fd793e6c40ad81404732b277d5eb449f8860", size = 4216972, upload-time = "2026-09-30T12:52:41.021Z" }, + { url = "https://files.pythonhosted.org/packages/1b/b2/533096223449c2e378e54157c090a349f230b67df3e45f8887897dcf2861/torchcodec-0.17.0-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:0ab2f00d849ba7b24e1ddcc4ed575d988fe4ed54f39c6a55575a1a47674eff9c", size = 8837707, upload-time = "2026-09-30T12:52:42.797Z" }, + { url = "https://files.pythonhosted.org/packages/74/00/9e004afc9dcee01e7daee1aa6e4a2d456b4cd401e11f65377d690f7cbde4/torchcodec-0.17.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:ba26d900b3941eac8176b5c42bed0e2315cc154f07037a385e211ec39af989bf", size = 9956598, upload-time = "2026-09-30T12:52:44.644Z" }, + { url = "https://files.pythonhosted.org/packages/c6/28/654778071dec72e39f98935486049e67b0e618079684b47a8d44e229cb9d/torchcodec-0.17.0-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:94f3391db6a35b1fe4f1c41f02e6fb3950ba62020f2812b67d317bc3d2cb8966", size = 4218648, upload-time = "2026-09-30T12:52:46.993Z" }, + { url = "https://files.pythonhosted.org/packages/ce/d1/d06d83bc55007dfbf70c885e5a58ca2d7224460197eb5320432068dbed5a/torchcodec-0.17.0-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b387f1d935d5a60ebad6de0144d728f5802de9253a65fb1570a6f2ad646c9d3b", size = 8843574, upload-time = "2026-09-30T12:52:48.584Z" }, + { url = "https://files.pythonhosted.org/packages/83/4b/46865454a4455368effee91baa51a1d1d1f7a9f3862b743a270c3b4a83e3/torchcodec-0.17.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:fc7021f10d38debd4fd47b4620da50d8cead981752bb359f7e3a615bbca2b7d3", size = 9965401, upload-time = "2026-09-30T12:52:50.848Z" }, + { url = "https://files.pythonhosted.org/packages/fa/eb/2fab706b1a8681e4d9281beeb3da93f7543f29a9fe0cb83850211cd28c7f/torchcodec-0.17.0-cp311-cp311-win_amd64.whl", hash = "sha256:9267dc578c4de6710d4686d52bb2e8dc9a44ad404e775957f4428375311ebc60", size = 6803717, upload-time = "2026-09-30T12:52:52.98Z" }, + { url = "https://files.pythonhosted.org/packages/e0/53/33acd60b2880ed4b0d6526436f691a0f447de7b53aceb42eb58a9719e474/torchcodec-0.17.0-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:3d69d30fbd2d5f9cc6d63c6d970acb5fa3aeddb5524a059f0ca538a83972f65c", size = 4219893, upload-time = "2026-09-30T12:52:54.86Z" }, + { url = "https://files.pythonhosted.org/packages/75/20/5692a1177b4be33086770970dcf83effa821376f447899cea3ba850ea2fa/torchcodec-0.17.0-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:81655cf25e168e74f27aee29db9fc16e37a9f5b2a188a0f7fe25cdfaab2ddd2e", size = 8854331, upload-time = "2026-09-30T12:52:56.445Z" }, + { url = "https://files.pythonhosted.org/packages/05/44/d00a4df9c8985c6ce9e6d1e3f861496cdcde8b5b4b7543358d49e03f7337/torchcodec-0.17.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:defd5f0901999c9e981f6465a0f4aac8b5a5147e38e539b77c0711c015cf5349", size = 9972897, upload-time = "2026-09-30T12:52:59.059Z" }, + { url = "https://files.pythonhosted.org/packages/f4/e4/0f2a356d6d08b1e2ece096288a02f2978806d077cac7a8340adf312b21ed/torchcodec-0.17.0-cp312-cp312-win_amd64.whl", hash = "sha256:0771bc114ff75a30f0ea8aa82cc7b259c518214eafe73fee2e606d745d77942b", size = 6810114, upload-time = "2026-09-30T12:53:00.966Z" }, + { url = "https://files.pythonhosted.org/packages/80/58/ff4d086c2b5e0807d1617b083eafcedebf8a9fbcabf2363ea8a92ad1c098/torchcodec-0.17.0-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:509558ec5d4f431fe2411d8f57e8e4fd6772f799bd44e570946b66989853c402", size = 4219907, upload-time = "2026-09-30T12:53:02.677Z" }, + { url = "https://files.pythonhosted.org/packages/8b/12/860dce5e7ef82879665ceeffd6f94e27fe51e390e66538205d8533311792/torchcodec-0.17.0-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b7a2a30f0372f4b0caae71a598c20e6d5ef7bf55dae2576003d6a6ffeccb9153", size = 8852117, upload-time = "2026-09-30T12:53:04.258Z" }, + { url = "https://files.pythonhosted.org/packages/68/2f/11ed8af589bcac10e6de4801728d21e44a71492578b6bb3d423841de2141/torchcodec-0.17.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:9eb27bc90ad1d60fbded1415e8106961d5f9e4bc80b3010217cfe6a08ad967ce", size = 9973287, upload-time = "2026-09-30T12:53:06.288Z" }, + { url = "https://files.pythonhosted.org/packages/cd/70/f0d7aa8e04ec12acfdb629f8c02a4bd3ec76186a6dc3002d2f6f1bdcb30f/torchcodec-0.17.0-cp313-cp313-win_amd64.whl", hash = "sha256:0af78216e6b2d13e33ecaa7b49bb06d6795cab39515449d40d2193cf1fa3e447", size = 6810219, upload-time = "2026-09-30T12:53:08.184Z" }, + { url = "https://files.pythonhosted.org/packages/c7/e0/78fd92b246cc0a9a0800306d82a743ec990105310595b68698061964574e/torchcodec-0.17.0-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:9567427ba28cee36704104f66bbaebfc210b89204fceb3063ca09a3fdeb0ac02", size = 4220040, upload-time = "2026-09-30T12:53:10.152Z" }, + { url = "https://files.pythonhosted.org/packages/01/36/75ba1e9bb6b226dce3e1cc2f6a1f589885e71aa4b831e96d1b00b9a000fd/torchcodec-0.17.0-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d9e682f91bc70a34a8c7ea3d302b3d6f52970b97c5b81aa098478197abf17fc6", size = 8853671, upload-time = "2026-09-30T12:53:11.75Z" }, + { url = "https://files.pythonhosted.org/packages/f7/98/986613945043b8beefd482c32c68fe20d7b9c79f28a5bd4da78a4144184e/torchcodec-0.17.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:aa5cf864c6a3ee42fa45961ec469ae359fbca448fb3440b86c84c9f77dd130aa", size = 9973805, upload-time = "2026-09-30T12:53:13.748Z" }, + { url = "https://files.pythonhosted.org/packages/86/59/c57bb24ffedebe1931be467371266234fb7292f462a4f65e0e4fe880ed9b/torchcodec-0.17.0-cp314-cp314-win_amd64.whl", hash = "sha256:916ff04f0accda465b6b33a9926d1b9daae6414e81abed5987b78c87374e6e8a", size = 6812058, upload-time = "2026-09-30T12:53:15.778Z" }, + { url = "https://files.pythonhosted.org/packages/a1/38/e27754be64b351b99d16e146c27fe1c8ef0abdd92186d33282afbf81801e/torchcodec-0.17.0-cp314-cp314t-macosx_14_0_arm64.whl", hash = "sha256:7c6004dbeb27a9e58bad0b3d10b4d7f7cda4407c0f10ba45a5a2ebd60aa33dbe", size = 4223965, upload-time = "2026-09-30T12:53:17.46Z" }, + { url = "https://files.pythonhosted.org/packages/3f/6e/838b73f42dcc8d892bba0b6f278f3ea7db577bce28b30b7306ae02224043/torchcodec-0.17.0-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8806e967d598d0d8ff0214837d9fb7565f06c7676f56ec172d5c57574e59d5b8", size = 8858951, upload-time = "2026-09-30T12:53:19.286Z" }, + { url = "https://files.pythonhosted.org/packages/a8/6a/3e1cad0310001fb50a222406f16c983c7bf1dbdbae73ae26c63288ddc308/torchcodec-0.17.0-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7d58e5211ed6f04aca2bd57e6a6104f9993ceb36a290db31de3f4e6096148f0b", size = 9975758, upload-time = "2026-09-30T12:53:21.471Z" }, +] + [[package]] name = "torchvision" version = "0.26.0" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and sys_platform != 'darwin'", + "python_full_version < '3.11' and sys_platform == 'darwin'", + "python_full_version < '3.11' and sys_platform != 'darwin'", +] dependencies = [ { name = "numpy" }, { name = "pillow" }, - { name = "torch" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" } }, ] wheels = [ { url = "https://files.pythonhosted.org/packages/74/b4/cdfee31e0402ea035135462cb0ab496e974d56fab6b4e7a1f0cbccb8cd28/torchvision-0.26.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:a06d4772a8e13e772906ed736cc53ec6639e5e60554f8e5fa6ca165aabebc464", size = 1863503, upload-time = "2026-03-23T18:13:01.384Z" }, @@ -9680,6 +11550,64 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/d0/6a/09f3844c10643f6c0de5d95abc863420cfaf194c88c7dffd0ac523e2015f/torchvision-0.26.0-cp314-cp314t-win_amd64.whl", hash = "sha256:e9d0e022c19a78552fb055d0414d47fecb4a649309b9968573daea160ba6869c", size = 4454275, upload-time = "2026-03-23T18:12:27.487Z" }, ] +[[package]] +name = "torchvision" +version = "0.28.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin'", +] +dependencies = [ + { name = "numpy", marker = "python_full_version >= '3.11'" }, + { name = "pillow", marker = "python_full_version >= '3.11'" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/b4/df/1ba039ad6cfe6e69209c36766b9b6e8c6fe92481c6d4e4ca52296f5f699d/torchvision-0.28.0-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:2a1ef4b6f4bf5828b48cfad97372c8982db906830884b2868ba5c3df937a7d81", size = 1856019, upload-time = "2026-07-08T16:07:59.283Z" }, + { url = "https://files.pythonhosted.org/packages/88/ea/5c70ecf86f8e95174a85061cea78683a7bb7f422f09c3f3d4f30b7600fa9/torchvision-0.28.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:546fd85345cf8652f6cd099d4f9884b0ca5c2f3fae78689a21dd2f35ea6b622f", size = 7838211, upload-time = "2026-07-08T16:07:27.023Z" }, + { url = "https://files.pythonhosted.org/packages/46/22/2f7ff1997d793e45d85fafa8374ee25348b7dae9ac521ba8751d7e1c75d5/torchvision-0.28.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:6dfb0f45e2b4ceb4e76f158c3fbb5f44387099f3c466e3423a09ab665a194aba", size = 7669419, upload-time = "2026-07-08T16:07:41.648Z" }, + { url = "https://files.pythonhosted.org/packages/42/d0/2b3c30834ff23acd3854d0ff59bc580711f4b36d725de40105a852ed3719/torchvision-0.28.0-cp310-cp310-win_amd64.whl", hash = "sha256:7fad44dc9582570c7d92c4487d36ac46998f40cc39b438e8b8f5111a935ce4e8", size = 3500355, upload-time = "2026-07-08T16:07:56.865Z" }, + { url = "https://files.pythonhosted.org/packages/7a/b2/1e010052079e4c577007b789db336ea7075f1a426e84d17121fbc3745516/torchvision-0.28.0-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:83fe6c020866a85acd7d97deccc45ff11d66daf42916d04396a4309c66c0ccb8", size = 1856017, upload-time = "2026-07-08T16:07:55.533Z" }, + { url = "https://files.pythonhosted.org/packages/27/be/1b9c5de9c655ca2df4a74100fa671a7b848532ff787e077ccde14a7dea2a/torchvision-0.28.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:5a38bc6da3d72621be003400b66f66a2b4c6d644fde05f680c2cb7ca8cf8dd6c", size = 7841822, upload-time = "2026-07-08T16:07:49.207Z" }, + { url = "https://files.pythonhosted.org/packages/0b/9b/f1e68e861d4462e3e195a642c2b448e7b7d3fad5f209487162b9a2133d9b/torchvision-0.28.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:7e80f543b22503d9415e126db5f0ff3917036925e38560ee6b9ae38c571a4002", size = 7670718, upload-time = "2026-07-08T16:07:46.525Z" }, + { url = "https://files.pythonhosted.org/packages/f5/de/1494610ff54cbb154beb55033cc2cd50f3de04dac132fa2dd00e4f2b2556/torchvision-0.28.0-cp311-cp311-win_amd64.whl", hash = "sha256:9a45ea67235d965ef52187130d20002a4de20c54ea3d927a24286961d268dc37", size = 3814319, upload-time = "2026-07-08T16:07:37.153Z" }, + { url = "https://files.pythonhosted.org/packages/15/49/c1cab1ecbb3ff1a380a3f99283db1dee61b8afe354f6352c643b65937130/torchvision-0.28.0-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:e9f54c30cd52e3ef7fd034cc69b7bb7e0964e1c8f8743e018ab92e95b40f9eee", size = 1856020, upload-time = "2026-07-08T16:07:52.182Z" }, + { url = "https://files.pythonhosted.org/packages/f0/4c/95233776e2def960e5abb7a07931230a545f43717a56a1e1140162033598/torchvision-0.28.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:5cf78ebc401ce64ae19b8c55de866bb836797d559a4de9c25ccbe74cfa642d3a", size = 7842127, upload-time = "2026-07-08T16:07:53.446Z" }, + { url = "https://files.pythonhosted.org/packages/93/e4/e9b2495d0d57b9f60d63c57d0a910410a81b4b073bf70917bef815291119/torchvision-0.28.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:028a3d481b37d785605620d7cdad897064c5a55bae2aa1f2658766333e291940", size = 7675040, upload-time = "2026-07-08T16:07:58.017Z" }, + { url = "https://files.pythonhosted.org/packages/7c/9c/55ed9cb6dfe3ee9c837df5cd0e758372e5829aa38b8dd71343aa632cc4e2/torchvision-0.28.0-cp312-cp312-win_amd64.whl", hash = "sha256:87dc16b2df427c1318ad335f1e2be2b3b15b2cf20f7934c83b0505a48425ee5d", size = 4085785, upload-time = "2026-07-08T16:07:50.928Z" }, + { url = "https://files.pythonhosted.org/packages/20/55/08a726c14c67b37c8aca04b077766909f1c7ed23f76116884fe63b9bd033/torchvision-0.28.0-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:d483b4aa3f5237569053f749cd1a2b5bb548ca456e40461a5dd087f21149d123", size = 1856021, upload-time = "2026-07-08T16:07:45.386Z" }, + { url = "https://files.pythonhosted.org/packages/db/8f/40beacd53809194f5259e590d1afaeaa8ad57da15f77c646e6560bcc4616/torchvision-0.28.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:bb6dd6918460ed89cc7644adcc2402991474d6933cf1ce92b390641cb233fddf", size = 7797014, upload-time = "2026-07-08T16:07:43.04Z" }, + { url = "https://files.pythonhosted.org/packages/32/db/062cdb5a84380a60439775311fff34d89229760d2a50680393dc18699956/torchvision-0.28.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:ad7b3a439265cc3739a4ab5b4c998c0e38ea99c0ee7ca4dea35c5d0b099ec237", size = 7674669, upload-time = "2026-07-08T16:07:38.91Z" }, + { url = "https://files.pythonhosted.org/packages/f3/a6/b4081e2d04e1541abf82785ac9e5178a494c19330391f551356c8c18b7b3/torchvision-0.28.0-cp313-cp313-win_amd64.whl", hash = "sha256:7e9dd6f60d6e15f8dc27d4f877fdb6002fc70d70272412135f1c2ff9cfa08d3b", size = 4157380, upload-time = "2026-07-08T16:07:40.22Z" }, + { url = "https://files.pythonhosted.org/packages/c5/b9/da40eca5bbe9596c12ae9899ab7abaf887f5e20f29d08b924b4633714821/torchvision-0.28.0-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:3bd9dba55224a9db4a2d77f6feaa5651770d8c8e86d3d0ddb0fa6bec54c8712b", size = 1856014, upload-time = "2026-07-08T16:07:44.282Z" }, + { url = "https://files.pythonhosted.org/packages/06/d6/313aafd3df4eaf5f330211bd4e75b7598bddbfee4f55580d3b58536e1b20/torchvision-0.28.0-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:89f90e29b0966352811b12589f3a3c61943bf2bb9487b9d7bbec10efb1096bb5", size = 7796873, upload-time = "2026-07-08T16:07:30.907Z" }, + { url = "https://files.pythonhosted.org/packages/b3/41/31f8e959ab8f942600b6357f8999c21d779d5fd3304b0fd204ff4b518239/torchvision-0.28.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:36beb0782976906069ca03d4c9aacaf4b6b838b06ed6c20960ea9c51cce7acdd", size = 7674634, upload-time = "2026-07-08T16:07:29.657Z" }, + { url = "https://files.pythonhosted.org/packages/15/15/4c5115253fd470672cdac0a1cf139e06b4f3e29d041238a2b255937f63be/torchvision-0.28.0-cp314-cp314-win_amd64.whl", hash = "sha256:3557cc7b539f46dabcda2b6f2b14017ccbeef024de466d4fc5835fc3f287f769", size = 4184005, upload-time = "2026-07-08T16:07:35.805Z" }, + { url = "https://files.pythonhosted.org/packages/6a/80/822a6163da716f8a78141cf6678d74e26a572285d4ea866ef8aa657bb307/torchvision-0.28.0-cp314-cp314t-macosx_14_0_arm64.whl", hash = "sha256:09ce8f56e81f19b9c378ae7bb109f83f6659fd8bc3cd14241a48e4af46e9ed49", size = 1856011, upload-time = "2026-07-08T16:07:33.404Z" }, + { url = "https://files.pythonhosted.org/packages/7f/d1/cd3f9463b39a790ec8c0c2f6e6c8061edb1562114d04fcdfa786ed889345/torchvision-0.28.0-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:62c7d110f86a039245b587e4fae60278c649f3bd42ff79cfbc1178eca4e72542", size = 7796742, upload-time = "2026-07-08T16:07:28.339Z" }, + { url = "https://files.pythonhosted.org/packages/d9/82/3e0a7ad18e99831e2d7f4713d3be717b7159ff5a920862dd5c23c454aa71/torchvision-0.28.0-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:904cf89af220f8c6b2ed0296bb5065b474ce43b77558e48b2bf9de8b0ba17204", size = 7675526, upload-time = "2026-07-08T16:07:34.572Z" }, + { url = "https://files.pythonhosted.org/packages/18/d4/23aea03b28297bc66a4461f55ae4296368a9d85fa9a454bafcb2a5348bd7/torchvision-0.28.0-cp314-cp314t-win_amd64.whl", hash = "sha256:46f581979c010ad6da6bd85ee602aa707e1ff44312670223b7a0ee517ad06d47", size = 4291452, upload-time = "2026-07-08T16:07:32.236Z" }, +] + [[package]] name = "tqdm" version = "4.67.3" @@ -9718,13 +11646,15 @@ resolution-markers = [ "python_full_version < '3.11' and sys_platform != 'darwin'", ] dependencies = [ - { name = "huggingface-hub", marker = "extra == 'extra-7-vec-inf-sglang'" }, + { name = "huggingface-hub", version = "1.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "huggingface-hub", version = "1.33.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "numpy", marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "packaging", marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "pyyaml", marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "regex", marker = "extra == 'extra-7-vec-inf-sglang'" }, - { name = "safetensors", marker = "extra == 'extra-7-vec-inf-sglang'" }, - { name = "tokenizers", marker = "extra == 'extra-7-vec-inf-sglang'" }, + { name = "safetensors", version = "0.7.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "safetensors", version = "0.8.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tokenizers", version = "0.22.2", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "tqdm", marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "typer", marker = "extra == 'extra-7-vec-inf-sglang'" }, ] @@ -9738,42 +11668,95 @@ name = "transformers" version = "5.9.0" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", ] dependencies = [ - { name = "huggingface-hub", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "numpy", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "packaging", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "pyyaml", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "regex", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "safetensors", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "tokenizers", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "tqdm", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "typer", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, + { name = "huggingface-hub", version = "1.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "numpy", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "packaging", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pyyaml", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "regex", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "safetensors", version = "0.7.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tokenizers", version = "0.22.2", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tqdm", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typer", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/51/58/7f843608f2e8421f86bb97060b54649be6239ec612b82bf9d41e65c26c00/transformers-5.9.0.tar.gz", hash = "sha256:25997cb8fa6053533171634b6162d7df54346530ec2aa9b42bb834e63668c842", size = 8642240, upload-time = "2026-05-20T14:50:49.278Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/02/ca/2eaa5359f2ccb8c2e1656bc26305ad0cf438aa392ce4b29ae67a315c186e/transformers-5.9.0-py3-none-any.whl", hash = "sha256:1d19509bcff7028ebc6b277d71caa712e8353778463d38764237d14b42b52788", size = 10787648, upload-time = "2026-05-20T14:50:45.337Z" }, ] +[[package]] +name = "transformers" +version = "5.19.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "huggingface-hub", version = "2.3.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "numpy", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "packaging", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pyyaml", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "regex", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "safetensors", version = "0.8.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tokenizers", version = "0.23.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "tqdm", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typer", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/65/4c/70012ed0795235fb6bd7806fafd4a1842f03e49ebbfa3bb9580e451948ce/transformers-5.19.0.tar.gz", hash = "sha256:87f38dd25e4521151b97e94520ac457f44a0ae8a8358a5b112daff6c64a822d6", size = 10032022, upload-time = "2026-10-06T16:39:00.229Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/b9/59/1e7f1212c215c73690ae6cab4678efe1555ff985953e145d88cc212859b0/transformers-5.19.0-py3-none-any.whl", hash = "sha256:afcd2dd5f603ed28c1e1fcb00a338ccbb4ef5f878ed289635df8b58187afb518", size = 12599123, upload-time = "2026-10-06T16:38:56.794Z" }, +] + [[package]] name = "triton" version = "3.6.0" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] wheels = [ { url = "https://files.pythonhosted.org/packages/44/ba/b1b04f4b291a3205d95ebd24465de0e5bf010a2df27a4e58a9b5f039d8f2/triton-3.6.0-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6c723cfb12f6842a0ae94ac307dba7e7a44741d720a40cf0e270ed4a4e3be781", size = 175972180, upload-time = "2026-01-20T16:15:53.664Z" }, { url = "https://files.pythonhosted.org/packages/8c/f7/f1c9d3424ab199ac53c2da567b859bcddbb9c9e7154805119f8bd95ec36f/triton-3.6.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a6550fae429e0667e397e5de64b332d1e5695b73650ee75a6146e2e902770bea", size = 188105201, upload-time = "2026-01-20T16:00:29.272Z" }, @@ -9791,13 +11774,54 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/f6/56/6113c23ff46c00aae423333eb58b3e60bdfe9179d542781955a5e1514cb3/triton-3.6.0-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:46bd1c1af4b6704e554cad2eeb3b0a6513a980d470ccfa63189737340c7746a7", size = 188397994, upload-time = "2026-01-20T16:01:14.236Z" }, ] +[[package]] +name = "triton" +version = "3.7.1" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +wheels = [ + { url = "https://files.pythonhosted.org/packages/ec/ea/629cc37436ca5df93ce98956d09cd2ca1498bfee8ef4972d2fe48b9f958c/triton-3.7.1-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:3daf64305d6cea88d3334c65ebc9bcd0c64c9564a977084366aa768d57cbcf64", size = 184551013, upload-time = "2026-06-17T20:03:37.551Z" }, + { url = "https://files.pythonhosted.org/packages/15/76/c79c34311625227a288df3e483fc5cdf3d596624cbd4b4758c4cbdc14af3/triton-3.7.1-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:ee89fbf782ec2ad50391dd1cf26cbea4f4467154c37f4773026da8fc31c0f58e", size = 197596267, upload-time = "2026-06-17T19:53:06.898Z" }, + { url = "https://files.pythonhosted.org/packages/7b/f9/19d842d06a08559534fa1eaab6ca551b1bcf40f06620bddec1babaa2772d/triton-3.7.1-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d4a0e1cd4c4a76370ed74a8432a53cea28716827d19e40ffc732233e35ceb3f6", size = 184664887, upload-time = "2026-06-17T20:03:42.913Z" }, + { url = "https://files.pythonhosted.org/packages/cd/5e/fce69606f7f240297f163e25539906732b199530d486ce67ae319877e821/triton-3.7.1-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:6744957e9fd610a29680ec2346057d0c86948ed3812468670719f391e94b44a5", size = 197701306, upload-time = "2026-06-17T19:53:13.673Z" }, + { url = "https://files.pythonhosted.org/packages/94/fa/f856e24deb462d5f18bd4b5a746957862ab9b6ee5834bda60605ec348366/triton-3.7.1-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:9497f2e696ee368862a181a90b2dcc03ca978cc4f602abd67c7d81022a6988e1", size = 184692359, upload-time = "2026-06-17T20:03:48.288Z" }, + { url = "https://files.pythonhosted.org/packages/c4/6f/fb96d15db6f36d6eae4cafb998c2e0353bf59d7c4ea1662d7497f269134a/triton-3.7.1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7e40869937a68206ec70d7f25bb7ec6433cb083f9135e1f36dbd318dc449a728", size = 197719725, upload-time = "2026-06-17T19:53:20.419Z" }, + { url = "https://files.pythonhosted.org/packages/00/42/c5089d4d9327fcd1e862c599cc2927f39418f84dd11a84cb2ccff9d4787a/triton-3.7.1-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:cdbfc09d9ec58bc5e68321525653220de7515c199e7a8097a97c85e62b52cd0a", size = 184694629, upload-time = "2026-06-17T20:03:53.444Z" }, + { url = "https://files.pythonhosted.org/packages/07/42/2c3ac59253ae8892b6f307875263dd23dc875cdf732d3aea40d6d41fb7cb/triton-3.7.1-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:58c0e131da05134a2a4788ccbcc0c1105cf0f54c8e98f19e34cd465396dc15eb", size = 197729241, upload-time = "2026-06-17T19:53:27.801Z" }, + { url = "https://files.pythonhosted.org/packages/40/71/e01aa7ad573883ed9456f130226babdec70b005e098c4d6226a6238e761b/triton-3.7.1-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:fe4ea396a06171f1f1f58cbd39c70b09294398f7dd7c620939bab54ad6f934fa", size = 184705764, upload-time = "2026-06-17T20:03:59.064Z" }, + { url = "https://files.pythonhosted.org/packages/a4/09/5683146fda6a2b569deb78ccfd8fbfea8bfe55f726b081c0a6bb18dd6f28/triton-3.7.1-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:2020153b08280415ec0da6607834e79166442147e78e144df06b508c75b186d2", size = 197729537, upload-time = "2026-06-17T19:53:35.516Z" }, + { url = "https://files.pythonhosted.org/packages/e9/f8/448220c3092019f9fdfab39ec47985968181d67da34b44f6a7f6280a5cbb/triton-3.7.1-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:c58e4c61f0c73b5dba3b5d19b4a7093c32f90dc18b2a7f121a7c16ccd31107b7", size = 184814760, upload-time = "2026-06-17T20:04:04.984Z" }, + { url = "https://files.pythonhosted.org/packages/f0/ac/229b7d4589d2e5937310e72c6d46e89599d16a4a12b479ffa1499fee8eb8/triton-3.7.1-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:10ba85fa2cca4a2fbdeb36bf1cb082f2c252bda55bf9fccd74f65ec5bc647e68", size = 197824404, upload-time = "2026-06-17T19:53:42.772Z" }, +] + +[[package]] +name = "truststore" +version = "0.10.5" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/ee/9f/c5201d42a484c061e528825fc8e2d565f5abd50a4ced6fb7d29c4ec99b2b/truststore-0.10.5.tar.gz", hash = "sha256:30d36967ccaded5cbb38d602c433f53600036c79d502f4533a49b60a03bbefcd", size = 28091, upload-time = "2026-10-12T22:27:31.808Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/51/e9/3a7820be2bb0fe53b6bc9c3be26d3d1158004e4c3ab953aa6840b955b1e9/truststore-0.10.5-py3-none-any.whl", hash = "sha256:9aaaedaefaf06d8b206278cf8b5012bc897f485a874503501e12d776df78951c", size = 19017, upload-time = "2026-10-12T22:27:30.377Z" }, +] + [[package]] name = "typer" version = "0.24.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "annotated-doc" }, - { name = "click" }, + { name = "click", version = "8.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "rich" }, { name = "shellingham" }, ] @@ -9811,23 +11835,33 @@ name = "typing-extensions" version = "4.15.0" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", @@ -9903,9 +11937,10 @@ name = "uvicorn" version = "0.44.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "click" }, - { name = "h11" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "click", version = "8.3.2", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "h11", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, + { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/5e/da/6eee1ff8b6cbeed47eeb5229749168e81eb4b7b999a1a15a7176e51410c9/uvicorn-0.44.0.tar.gz", hash = "sha256:6c942071b68f07e178264b9152f1f16dfac5da85880c4ce06366a96d70d4f31e", size = 86947, upload-time = "2026-04-06T09:23:22.826Z" } wheels = [ @@ -9914,13 +11949,13 @@ wheels = [ [package.optional-dependencies] standard = [ - { name = "colorama", marker = "sys_platform == 'win32'" }, - { name = "httptools" }, - { name = "python-dotenv" }, - { name = "pyyaml" }, - { name = "uvloop", marker = "platform_python_implementation != 'PyPy' and sys_platform != 'cygwin' and sys_platform != 'win32'" }, - { name = "watchfiles" }, - { name = "websockets" }, + { name = "colorama", marker = "python_full_version >= '3.11' and sys_platform == 'win32'" }, + { name = "httptools", marker = "python_full_version >= '3.11'" }, + { name = "python-dotenv", marker = "python_full_version >= '3.11'" }, + { name = "pyyaml", marker = "python_full_version >= '3.11'" }, + { name = "uvloop", marker = "python_full_version >= '3.11' and platform_python_implementation != 'PyPy' and sys_platform != 'cygwin' and sys_platform != 'win32'" }, + { name = "watchfiles", marker = "python_full_version >= '3.11'" }, + { name = "websockets", marker = "python_full_version >= '3.11'" }, ] [[package]] @@ -9972,7 +12007,8 @@ name = "vec-inf" version = "0.9.0" source = { editable = "." } dependencies = [ - { name = "click" }, + { name = "click", version = "8.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.12'" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.12'" }, { name = "pyyaml" }, @@ -9989,8 +12025,8 @@ sglang = [ ] vllm = [ { name = "ray", extra = ["default"], marker = "extra == 'extra-7-vec-inf-vllm'" }, - { name = "torchcodec", version = "0.9.1", source = { registry = "https://pypi.org/simple" } }, - { name = "vllm", extra = ["audio", "bench"], marker = "extra == 'extra-7-vec-inf-vllm'" }, + { name = "torchcodec", version = "0.17.0", source = { registry = "https://pypi.org/simple" } }, + { name = "vllm", extra = ["audio", "bench"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] [package.dev-dependencies] @@ -10021,12 +12057,16 @@ inference = [ { name = "ax-platform", version = "1.2.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "cupy-cuda12x" }, { name = "flashinfer-python", version = "0.6.11.post1", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, - { name = "flashinfer-python", version = "0.6.11.post2", source = { registry = "https://pypi.org/simple" }, marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, + { name = "flashinfer-python", version = "0.6.11.post2", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "flashinfer-python", version = "0.7.0.post1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "py3nvml" }, - { name = "torch" }, - { name = "wandb" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11' or extra == 'extra-7-vec-inf-sglang'" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "wandb", version = "0.25.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11' or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "wandb", version = "0.31.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "xgrammar", version = "0.2.0", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, - { name = "xgrammar", version = "0.2.1", source = { registry = "https://pypi.org/simple" }, marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, + { name = "xgrammar", version = "0.2.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "xgrammar", version = "0.2.8", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] [package.metadata] @@ -10041,10 +12081,8 @@ requires-dist = [ { name = "sgl-kernel", marker = "extra == 'sglang'", specifier = ">=0.3.0" }, { name = "sglang", marker = "extra == 'sglang'", specifier = ">=0.5.5" }, { name = "torchao", marker = "extra == 'sglang'", specifier = ">=0.9.0" }, - { name = "torchcodec", marker = "extra == 'vllm'", specifier = ">=0.9.0,<0.10.0" }, - { name = "vllm", marker = "extra == 'vllm'", specifier = ">=0.11.2" }, - { name = "vllm", extras = ["audio"], marker = "extra == 'vllm'" }, - { name = "vllm", extras = ["bench"], marker = "extra == 'vllm'" }, + { name = "torchcodec", marker = "extra == 'vllm'", specifier = ">=0.14.0" }, + { name = "vllm", extras = ["audio", "bench"], marker = "python_full_version >= '3.11' and extra == 'vllm'", specifier = "==0.32.0" }, ] provides-extras = ["vllm", "sglang"] @@ -10109,124 +12147,144 @@ wheels = [ [[package]] name = "vllm" -version = "0.22.0" -source = { registry = "https://pypi.org/simple" } -dependencies = [ - { name = "aiohttp" }, - { name = "anthropic" }, - { name = "apache-tvm-ffi", version = "0.1.9", source = { registry = "https://pypi.org/simple" } }, - { name = "blake3" }, - { name = "cachetools" }, - { name = "cbor2" }, - { name = "cloudpickle" }, - { name = "compressed-tensors" }, - { name = "depyf" }, - { name = "diskcache" }, - { name = "einops" }, - { name = "fastapi", extra = ["standard"], marker = "extra == 'extra-7-vec-inf-vllm'" }, - { name = "fastsafetensors" }, - { name = "filelock", version = "4.1.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "filelock", version = "4.2.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "flashinfer-cubin", version = "0.6.11.post2", source = { registry = "https://pypi.org/simple" } }, - { name = "flashinfer-python", version = "0.6.11.post2", source = { registry = "https://pypi.org/simple" } }, - { name = "gguf" }, - { name = "humming-kernels", extra = ["cu13"], marker = "extra == 'extra-7-vec-inf-vllm'" }, - { name = "ijson" }, - { name = "lark" }, - { name = "llguidance", version = "1.7.6", source = { registry = "https://pypi.org/simple" }, marker = "platform_machine == 'aarch64' or platform_machine == 'arm64' or platform_machine == 'ppc64le' or platform_machine == 'x86_64'" }, - { name = "lm-format-enforcer" }, - { name = "mcp" }, - { name = "mistral-common", extra = ["image"], marker = "extra == 'extra-7-vec-inf-vllm'" }, - { name = "model-hosting-container-standards" }, - { name = "msgspec" }, - { name = "ninja" }, - { name = "numba" }, - { name = "numpy" }, - { name = "nvidia-cudnn-frontend" }, - { name = "nvidia-cutlass-dsl", version = "4.5.2", source = { registry = "https://pypi.org/simple" }, extra = ["cu13"], marker = "extra == 'extra-7-vec-inf-vllm'" }, - { name = "openai", version = "2.31.0", source = { registry = "https://pypi.org/simple" } }, - { name = "openai-harmony", version = "0.0.8", source = { registry = "https://pypi.org/simple" } }, - { name = "opencv-python-headless" }, - { name = "opentelemetry-api" }, - { name = "opentelemetry-exporter-otlp" }, - { name = "opentelemetry-sdk" }, - { name = "opentelemetry-semantic-conventions-ai" }, - { name = "outlines-core", version = "0.2.14", source = { registry = "https://pypi.org/simple" } }, - { name = "partial-json-parser" }, - { name = "pillow" }, - { name = "prometheus-client" }, - { name = "prometheus-fastapi-instrumentator" }, - { name = "protobuf" }, - { name = "psutil" }, - { name = "py-cpuinfo" }, - { name = "pybase64" }, - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +version = "0.32.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "aiohttp", marker = "python_full_version >= '3.11'" }, + { name = "anthropic", marker = "python_full_version >= '3.11'" }, + { name = "apache-tvm-ffi", version = "0.1.11", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "blake3", marker = "python_full_version >= '3.11'" }, + { name = "cachetools", marker = "python_full_version >= '3.11'" }, + { name = "cbor2", marker = "python_full_version >= '3.11'" }, + { name = "cloudpickle", marker = "python_full_version >= '3.11'" }, + { name = "compressed-tensors", version = "0.17.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "depyf", marker = "python_full_version >= '3.11'" }, + { name = "einops", marker = "python_full_version >= '3.11'" }, + { name = "fastapi", extra = ["standard"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "fastsafetensors", marker = "python_full_version >= '3.11'" }, + { name = "filelock", version = "4.2.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "flashinfer-python", version = "0.7.0.post1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "huggingface-hub", version = "2.3.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "humming-kernels", extra = ["cu13"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "ijson", marker = "python_full_version >= '3.11'" }, + { name = "instanttensor", marker = "python_full_version >= '3.11'" }, + { name = "jsonschema", marker = "python_full_version >= '3.11'" }, + { name = "lark", marker = "python_full_version >= '3.11'" }, + { name = "llguidance", version = "1.7.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 'aarch64') or (python_full_version >= '3.11' and platform_machine == 'arm64') or (python_full_version >= '3.11' and platform_machine == 'ppc64le') or (python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and platform_machine == 'x86_64')" }, + { name = "lm-format-enforcer", marker = "python_full_version >= '3.11'" }, + { name = "mcp", marker = "python_full_version >= '3.11'" }, + { name = "mistral-common", version = "1.12.0", source = { registry = "https://pypi.org/simple" }, extra = ["image"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "model-hosting-container-standards", marker = "python_full_version >= '3.11'" }, + { name = "msgspec", marker = "python_full_version >= '3.11'" }, + { name = "ninja", marker = "python_full_version >= '3.11'" }, + { name = "numba", marker = "python_full_version >= '3.11'" }, + { name = "numpy", marker = "python_full_version >= '3.11'" }, + { name = "nvidia-cudnn-frontend", version = "1.31.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "nvidia-cutlass-dsl", version = "4.8.0", source = { registry = "https://pypi.org/simple" }, extra = ["cu13"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "nvtx", marker = "python_full_version >= '3.11'" }, + { name = "openai", version = "2.31.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "opencv-python-headless", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-api", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-exporter-otlp", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-sdk", marker = "python_full_version >= '3.11'" }, + { name = "opentelemetry-semantic-conventions-ai", marker = "python_full_version >= '3.11'" }, + { name = "oss-harmony", marker = "python_full_version >= '3.11'" }, + { name = "outlines-core", version = "0.2.14", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "partial-json-parser", marker = "python_full_version >= '3.11'" }, + { name = "pillow", marker = "python_full_version >= '3.11'" }, + { name = "prometheus-client", marker = "python_full_version >= '3.11'" }, + { name = "prometheus-fastapi-instrumentator", marker = "python_full_version >= '3.11'" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "psutil", marker = "python_full_version >= '3.11'" }, + { name = "py-cpuinfo", marker = "python_full_version >= '3.11'" }, + { name = "pybase64", marker = "python_full_version >= '3.11'" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "python-json-logger" }, - { name = "pyyaml" }, - { name = "pyzmq" }, - { name = "quack-kernels" }, - { name = "regex" }, - { name = "requests" }, - { name = "safetensors" }, - { name = "sentencepiece" }, - { name = "setproctitle" }, + { name = "pynvvideocodec", marker = "python_full_version >= '3.11'" }, + { name = "python-json-logger", marker = "python_full_version >= '3.11'" }, + { name = "pyyaml", marker = "python_full_version >= '3.11'" }, + { name = "pyzmq", marker = "python_full_version >= '3.11'" }, + { name = "quack-kernels", version = "0.6.5", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "regex", marker = "python_full_version >= '3.11'" }, + { name = "requests", marker = "python_full_version >= '3.11'" }, + { name = "safetensors", version = "0.8.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "sentencepiece", marker = "python_full_version >= '3.11'" }, + { name = "setproctitle", marker = "python_full_version >= '3.11'" }, { name = "setuptools", marker = "python_full_version >= '3.12'" }, { name = "six", marker = "python_full_version >= '3.12'" }, - { name = "tiktoken" }, - { name = "tilelang", version = "0.1.9", source = { registry = "https://pypi.org/simple" } }, - { name = "tokenizers" }, - { name = "tokenspeed-mla", version = "0.1.2", source = { registry = "https://pypi.org/simple" } }, - { name = "torch" }, - { name = "torchaudio" }, - { name = "torchvision" }, - { name = "tqdm" }, - { name = "transformers", version = "5.9.0", source = { registry = "https://pypi.org/simple" } }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "watchfiles" }, - { name = "xgrammar", version = "0.2.1", source = { registry = "https://pypi.org/simple" }, marker = "platform_machine == 'aarch64' or platform_machine == 'arm64' or platform_machine == 'ppc64le' or platform_machine == 's390x' or platform_machine == 'x86_64'" }, + { name = "starlette", marker = "python_full_version >= '3.11'" }, + { name = "tiktoken", marker = "python_full_version >= '3.11'" }, + { name = "tilelang", version = "0.1.12", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "tokenizers", version = "0.23.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "tokenspeed-mla", version = "0.2.15", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11' and sys_platform == 'linux'" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "torchaudio", marker = "python_full_version >= '3.11'" }, + { name = "torchcodec", version = "0.17.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "torchvision", version = "0.28.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "tqdm", marker = "python_full_version >= '3.11'" }, + { name = "transformers", version = "5.19.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "watchfiles", marker = "python_full_version >= '3.11'" }, + { name = "xgrammar", version = "0.2.8", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 'aarch64') or (python_full_version >= '3.11' and platform_machine == 'arm64') or (python_full_version >= '3.11' and platform_machine == 'ppc64le') or (python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and platform_machine == 'x86_64')" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/e2/bf/46631fd8e2e9d81c5abe2ab923e5367754bc0cad685c4ddac1d5d86d91b5/vllm-0.22.0.tar.gz", hash = "sha256:6d41581a9e5288cd69278518a550c6d7ce510ae27a506556a3427d01284be7fe", size = 36239170, upload-time = "2026-05-29T10:35:39.448Z" } +sdist = { url = "https://files.pythonhosted.org/packages/d5/df/5471312c76e335eaa2b8333e87ea39f0d37bfe24f10a582620e34ff12e33/vllm-0.32.0.tar.gz", hash = "sha256:d2c957d5eee3d0985012151315ddc62cbe0cd368fc96a5fc3510438d9dddf2ad", size = 44549956, upload-time = "2026-10-18T10:04:07.069Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/ed/e8/05a69dbd7416c5a5ac91f51e626fede9ceeabe9c6fe243fc11e2b3e1ad3e/vllm-0.22.0-cp38-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:0fbe1ff32e9ad82c56b002de11b061ca6b5b8a256cd11473946d2222115ed267", size = 252942448, upload-time = "2026-05-29T10:32:10.261Z" }, - { url = "https://files.pythonhosted.org/packages/ca/23/3f7f759763fb9b4cf5787bcb4a43f74904f8e644d53d4fdb4e19654a92fd/vllm-0.22.0-cp38-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:c387a977e35795e8f77b009e019e69722963819c26b55e4a679e09d4279ae35d", size = 261034920, upload-time = "2026-05-29T10:33:10.357Z" }, + { url = "https://files.pythonhosted.org/packages/02/7b/19ee98705e4a357fdb6158083c4aedb93bf548bf89fa827023449e22c98d/vllm-0.32.0-cp38-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:d0620dbdf99528bcfebe551a8906347a229921fb1602f887bd1286a920da2d44", size = 321437289, upload-time = "2026-10-18T10:03:29.256Z" }, + { url = "https://files.pythonhosted.org/packages/a5/0d/0b15fda9a56ac485ebb607e466f0e3386c8c115bafeced6514298bd325b5/vllm-0.32.0-cp38-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:5e2ecdc5b0baf72921c288486417ca82778456a448e4fe60b62ee43c0dfc8611", size = 326518482, upload-time = "2026-10-18T10:03:52.559Z" }, ] [package.optional-dependencies] audio = [ - { name = "av" }, - { name = "mistral-common", extra = ["audio"], marker = "extra == 'extra-7-vec-inf-vllm'" }, - { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "soundfile" }, + { name = "av", marker = "python_full_version >= '3.11'" }, + { name = "mistral-common", version = "1.12.0", source = { registry = "https://pypi.org/simple" }, extra = ["audio"], marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "soundfile", marker = "python_full_version >= '3.11'" }, + { name = "soxr", marker = "python_full_version >= '3.11'" }, ] bench = [ - { name = "datasets" }, - { name = "matplotlib" }, - { name = "pandas" }, - { name = "plotly" }, - { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "seaborn" }, + { name = "datasets", version = "5.1.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "matplotlib", marker = "python_full_version >= '3.11'" }, + { name = "pandas", marker = "python_full_version >= '3.11'" }, + { name = "plotly", marker = "python_full_version >= '3.11'" }, + { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "seaborn", marker = "python_full_version >= '3.11'" }, ] [[package]] name = "wandb" version = "0.25.1" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] dependencies = [ - { name = "click" }, - { name = "gitpython" }, - { name = "packaging" }, - { name = "platformdirs" }, - { name = "protobuf" }, - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.12'" }, - { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.12'" }, - { name = "pyyaml" }, - { name = "requests" }, - { name = "sentry-sdk" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "click", version = "8.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "gitpython", marker = "python_full_version < '3.11' or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "packaging", marker = "python_full_version < '3.11' or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "platformdirs", marker = "python_full_version < '3.11' or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "6.33.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11' or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11' or (python_full_version == '3.11.*' and platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pyyaml", marker = "python_full_version < '3.11' or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "requests", marker = "python_full_version < '3.11' or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "sentry-sdk", marker = "python_full_version < '3.11' or (platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine != 's390x' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/60/bb/eb579bf9abac70934a014a9d4e45346aab307994f3021d201bebe5fa25ec/wandb-0.25.1.tar.gz", hash = "sha256:b2a95cd777ecbe7499599a43158834983448a0048329bc7210ef46ca18d21994", size = 43983308, upload-time = "2026-03-10T23:51:44.227Z" } @@ -10242,6 +12300,60 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/c4/e8/76836b75d401ff5912aaf513176e64557ceaec4c4946bfd38a698ff84d48/wandb-0.25.1-py3-none-win_arm64.whl", hash = "sha256:cc7c34b70cf4b7be4d395541e82e325fd9d2be978d62c9ec01f1a7141523b6bb", size = 22080774, upload-time = "2026-03-10T23:51:40.196Z" }, ] +[[package]] +name = "wandb" +version = "0.31.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra == 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "click", version = "8.5.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "cwevals", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "opentelemetry-api", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "opentelemetry-exporter-otlp-proto-http", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "opentelemetry-sdk", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "packaging", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "platformdirs", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "protobuf", version = "7.36.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and platform_machine == 's390x') or (python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and platform_machine == 's390x') or (python_full_version >= '3.12' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pyyaml", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "requests", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "xxhash", marker = "(python_full_version >= '3.11' and platform_machine == 's390x') or (python_full_version >= '3.11' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/07/a5/8fc1c319f2db9be5a707c25a76ecba1b2b9450a44a05b295ff5aa379bbfa/wandb-0.31.0.tar.gz", hash = "sha256:dcdaa635576ff17b18a757b39fc22b0db3d1c6e01d5ce090fe80f6adc776028d", size = 40420416, upload-time = "2026-10-16T18:56:13.817Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/f2/62/ee7f6e5133b68a5b3e389ee9684079b018af56fc2a20200c03b66c1dc226/wandb-0.31.0-py3-none-macosx_13_0_arm64.whl", hash = "sha256:28389f1c932b5d3fc11293fd50c24f308c6a1e1dd952783b96d9703062302bb5", size = 18355517, upload-time = "2026-10-16T18:55:55.879Z" }, + { url = "https://files.pythonhosted.org/packages/15/3b/8b4f8e09e63e9c67146f35c6da5d78cdce89eb193385ca990e8aaa5681c7/wandb-0.31.0-py3-none-macosx_13_0_x86_64.whl", hash = "sha256:654ef6d854516888091aff582a243714b02cefb09c4a0e1186cc6a7c9f93de50", size = 19882635, upload-time = "2026-10-16T18:55:58.469Z" }, + { url = "https://files.pythonhosted.org/packages/e3/c8/aa5e7ab3d947928d8a41e676d519e3fd37e7bf467ae72261fcb0823199db/wandb-0.31.0-py3-none-manylinux_2_28_aarch64.whl", hash = "sha256:0cbd8986ca56115897defcfd8ad8def1a871b99d1e8bde4f106633dc9969b7a3", size = 18532675, upload-time = "2026-10-16T18:56:00.827Z" }, + { url = "https://files.pythonhosted.org/packages/2a/07/5594faeaea9440be2ea01cee692573416b3997879df67f0e940bfbfc2d26/wandb-0.31.0-py3-none-manylinux_2_28_x86_64.whl", hash = "sha256:009bcf370598d5e7a3d3bb66e33510f78c17fcc0505f8aa7ecf51dbc4378f979", size = 20337507, upload-time = "2026-10-16T18:56:02.937Z" }, + { url = "https://files.pythonhosted.org/packages/58/06/680fce9a6857a1376cab5fecf357409c27ae4a98908e3936a256487eaa39/wandb-0.31.0-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:cb20ed655fb6e301b27f9974ae639176ec3a0e81aa9135d107491d618206ad5f", size = 18637167, upload-time = "2026-10-16T18:56:05.129Z" }, + { url = "https://files.pythonhosted.org/packages/b6/0a/278fdc0edd079d01d52f4bb7cf9620ca6ec86c5b80d22fb41a8421d6bc05/wandb-0.31.0-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:df16a46cd6468e6580783b02560c79991c181f63a4691949020fff1b4ac87072", size = 20520389, upload-time = "2026-10-16T18:56:07.402Z" }, + { url = "https://files.pythonhosted.org/packages/90/7a/aaee2e55d1aa6f855d01e702263e2e0ef733da45583ecd9836d68ea3d67d/wandb-0.31.0-py3-none-win_amd64.whl", hash = "sha256:488f8ef63cae217ba220e6d8c163ee8c5b98320ee1866ce774c4da0847235f02", size = 19596112, upload-time = "2026-10-16T18:56:09.473Z" }, + { url = "https://files.pythonhosted.org/packages/75/d8/03efc83dccada300944fce5ac0bac650835855fdfed9654833c9e7c29f81/wandb-0.31.0-py3-none-win_arm64.whl", hash = "sha256:d155662ab444ab0cd12afa87943a147193e9b9301134b0b34e8c97c68a4bbba5", size = 17972736, upload-time = "2026-10-16T18:56:11.488Z" }, +] + [[package]] name = "watchdog" version = "6.0.0" @@ -10279,7 +12391,7 @@ name = "watchfiles" version = "1.1.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "anyio" }, + { name = "anyio", marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or extra == 'extra-7-vec-inf-sglang'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/c2/c9/8869df9b2a2d6c59d79220a4db37679e74f807c559ffe5265e08b227a210/watchfiles-1.1.1.tar.gz", hash = "sha256:a173cb5c16c4f40ab19cecf48a534c409f7ea983ab8fed0741304a1c0a31b3f2", size = 94440, upload-time = "2025-10-14T15:06:21.08Z" } wheels = [ @@ -10579,9 +12691,9 @@ dependencies = [ { name = "numpy", marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "torch", marker = "extra == 'extra-7-vec-inf-sglang'" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, { name = "transformers", version = "5.6.0", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-sglang'" }, - { name = "triton", marker = "(platform_machine == 'x86_64' and sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "triton", version = "3.6.0", source = { registry = "https://pypi.org/simple" }, marker = "(platform_machine == 'x86_64' and sys_platform == 'linux' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] @@ -10623,32 +12735,17 @@ name = "xgrammar" version = "0.2.1" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.14' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.14' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.13.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.12.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", - "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", "python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", ] dependencies = [ - { name = "apache-tvm-ffi", version = "0.1.9", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-vec-inf-vllm'" }, - { name = "apache-tvm-ffi", version = "0.1.10", source = { registry = "https://pypi.org/simple" }, marker = "(extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "numpy", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.12' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "torch", marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "transformers", version = "5.9.0", source = { registry = "https://pypi.org/simple" }, marker = "extra != 'extra-7-vec-inf-sglang' or extra == 'extra-7-vec-inf-vllm'" }, - { name = "triton", marker = "(platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, - { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "apache-tvm-ffi", version = "0.1.10", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "numpy", marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "torch", version = "2.11.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "transformers", version = "5.9.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "triton", version = "3.6.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "typing-extensions", version = "4.16.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, ] sdist = { url = "https://files.pythonhosted.org/packages/d8/ea/6394caddd078d33772070eefaaf77cb0a826a3047b908b688dece7d040b5/xgrammar-0.2.1.tar.gz", hash = "sha256:4c48c251b75d211e9ffa7f4f4ac8b5b0164f89fd5f0d1883ad7ff4554922030d", size = 2427065, upload-time = "2026-05-17T21:39:26.576Z" } @@ -10684,6 +12781,83 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/1b/d1/acb2d68930535364453669af1a3b0f3f95f647f4bddb9ea20c5fab295672/xgrammar-0.2.1-cp314-cp314t-win_amd64.whl", hash = "sha256:743196d955b82181af265c358a84b9e1e80ea0e368f129e560c898555a8df9f6", size = 7504318, upload-time = "2026-05-17T21:39:03.528Z" }, ] +[[package]] +name = "xgrammar" +version = "0.2.8" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'emscripten' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine == 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.14' and platform_machine != 's390x' and sys_platform != 'darwin' and sys_platform != 'emscripten' and sys_platform != 'win32' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.13.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.12.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform == 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine == 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and platform_machine != 's390x' and sys_platform != 'darwin' and extra != 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm'", + "python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang' and extra != 'extra-7-vec-inf-vllm'", +] +dependencies = [ + { name = "apache-tvm-ffi", version = "0.1.11", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "numpy", marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.12.5", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version == '3.11.*' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "pydantic", version = "2.13.4", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.12' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "torch", version = "2.13.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "transformers", version = "5.19.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "triton", version = "3.7.1", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and platform_machine == 'x86_64' and sys_platform == 'linux' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "typing-extensions", version = "4.15.0", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra != 'extra-7-vec-inf-sglang') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/03/0f/29f521535729be740779c2c98b630ef6607cf28be9b669895dd2b821ff63/xgrammar-0.2.8.tar.gz", hash = "sha256:8305646fd2b319d6c3003f51a1665e54cbbe9cba1fb636bc41d0b46434ccbb80", size = 2689558, upload-time = "2026-09-24T18:35:21.261Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/46/60/9973355ae43d92f7e7057b0fbdba78a24fc8350436bdb4148b0fd9ecf297/xgrammar-0.2.8-cp310-cp310-macosx_10_14_x86_64.whl", hash = "sha256:9c04f0dafb88b7f55559463a40ad7147f7d62c0a547f7c807054627dc6ed992b", size = 1496691, upload-time = "2026-09-24T18:34:12.098Z" }, + { url = "https://files.pythonhosted.org/packages/d1/80/f901045de56c377a6a426992adab4f5fbe9ee27b5fc05d0e9960f8c58c38/xgrammar-0.2.8-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:467533c9998566b9bda6889834a7e200cd75228e6b453d8a42abd3aa86158512", size = 1339886, upload-time = "2026-09-24T18:34:13.913Z" }, + { url = "https://files.pythonhosted.org/packages/de/5b/fdc88db44bf413e5d78802bfcd7b73f4535a03b14b5a26224a14387301e3/xgrammar-0.2.8-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:ae437044b736797a3a0ad03a06ab5d156903b350f7f0809591f47e0571e27e51", size = 1157373, upload-time = "2026-09-24T18:34:15.704Z" }, + { url = "https://files.pythonhosted.org/packages/c8/3d/133d4bf964e5bcba1314d1fd0b8daaf59ad010b97f02333c31e940918b96/xgrammar-0.2.8-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:d6576539ade4b2404c38f606f4d04a1d6d22023fa03f9cab98bf894877696024", size = 1311291, upload-time = "2026-09-24T18:34:17.094Z" }, + { url = "https://files.pythonhosted.org/packages/4f/07/6be2505a0920fdec02ee65586ea68eb910a50dad3c7e096e9526880bfd1c/xgrammar-0.2.8-cp310-cp310-win_amd64.whl", hash = "sha256:1a1dba06ff9f8436163ccc688ea0b24e790d99262f628ac4fc46c4fc09e18630", size = 1115211, upload-time = "2026-09-24T18:34:19.312Z" }, + { url = "https://files.pythonhosted.org/packages/72/b4/23d85194c3445f00b747cf276310e97f3eaab346767780a1cf48c940e62c/xgrammar-0.2.8-cp310-cp310-win_arm64.whl", hash = "sha256:1fe8b3889a9079bc3fbe9187941c7d902da7b9cb2dd915b2a004fa5a5922bcc5", size = 1084231, upload-time = "2026-09-24T18:34:21.02Z" }, + { url = "https://files.pythonhosted.org/packages/48/ae/44fe49b3defa896f22f1ee7c4e935a17690e0711db57733ef2add5ce627b/xgrammar-0.2.8-cp311-cp311-macosx_10_14_x86_64.whl", hash = "sha256:f56156ca0abdaf089c3f87fcfa7bafeab47df3374f221b67b2e4b4be8c819125", size = 1496691, upload-time = "2026-09-24T18:34:22.692Z" }, + { url = "https://files.pythonhosted.org/packages/f2/eb/de42bf71f53b1b6a1e3077740fb2548108bffe7896a9fe2833d4b4c825d6/xgrammar-0.2.8-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:d125a174b609d9d540dcf12dadf9030422597c175f0493ea9ff27dfeae025664", size = 1339888, upload-time = "2026-09-24T18:34:24.259Z" }, + { url = "https://files.pythonhosted.org/packages/b2/5a/721a753300c7eb66724e4f1f23657b9537ffc5e41aff12816228e337a8a2/xgrammar-0.2.8-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:738d967f874dd916ddb73b39e533caf7254b5b81d579f44eddbe205cc59dcee1", size = 1157368, upload-time = "2026-09-24T18:34:26.084Z" }, + { url = "https://files.pythonhosted.org/packages/21/14/d87e336c1b4ed0109b7250cdc102958c1cd3ac271d0b5b34fdb46951f730/xgrammar-0.2.8-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:885d7468636a8ca829513b9740fef21cbb438239fa3f91cce06e95e31973b8c6", size = 1311297, upload-time = "2026-09-24T18:34:27.551Z" }, + { url = "https://files.pythonhosted.org/packages/d8/85/f8c99af8ff77b57f6d3fbed124164853bb17c43fd4b8e0afc9fe0d813370/xgrammar-0.2.8-cp311-cp311-win_amd64.whl", hash = "sha256:155091c54e3cd5cde6c29e69339a74e5c95da83bb0294acd2abb9a70f85f674e", size = 1115214, upload-time = "2026-09-24T18:34:29.101Z" }, + { url = "https://files.pythonhosted.org/packages/79/44/2a986cdeffad64925583a52f0ac6251146b4af54f86df205924b24dd2de3/xgrammar-0.2.8-cp311-cp311-win_arm64.whl", hash = "sha256:50f8bc56adc2fccfe008281435ebef30d066977b611652cf51d0b57b16f48ec5", size = 1084232, upload-time = "2026-09-24T18:34:31.002Z" }, + { url = "https://files.pythonhosted.org/packages/3b/70/93ac6205161d7eb53ca82e51a51f2f2eea55d51994c3d4e841e12a517da5/xgrammar-0.2.8-cp312-cp312-macosx_10_14_x86_64.whl", hash = "sha256:9b4fb0907e2ae7b32e908e9da97e5410ee1f792d802031a46e4292f2ebe46c33", size = 1496688, upload-time = "2026-09-24T18:34:32.896Z" }, + { url = "https://files.pythonhosted.org/packages/b6/b8/44cf923b934efc3b4e812279c85777ba6a8977f6ce0f3f769bead42816d2/xgrammar-0.2.8-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:e01f70fbb788d5a95e6a2ac686b746c2dd03bffc07c7d94f10ec3ca2a95c4447", size = 1339886, upload-time = "2026-09-24T18:34:34.612Z" }, + { url = "https://files.pythonhosted.org/packages/b1/74/cdb56cc5afde708c6c39ed3d80680d7af6d12398bbb2681268bd5ac67eb7/xgrammar-0.2.8-cp312-cp312-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:97732b6e5f11f21fd98a6cd9525241fc5f9d92f415329200a9adf8fc7453d8ec", size = 1157369, upload-time = "2026-09-24T18:34:36.456Z" }, + { url = "https://files.pythonhosted.org/packages/2c/2e/bab212484b84c296974168021693278f8d73d87ebccc4a6cc04e5ae2a061/xgrammar-0.2.8-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4045b096735bb2523e0d874039a55994c1b390620d547c4a0d4f43b9a106177b", size = 1311291, upload-time = "2026-09-24T18:34:38.217Z" }, + { url = "https://files.pythonhosted.org/packages/c2/e2/6157eab393b7963f0933dc22c222b0667407ce230eac26bb40645a28748f/xgrammar-0.2.8-cp312-cp312-win_amd64.whl", hash = "sha256:dbcf6718e049cbe429742cdb05f85c48135c2e389d45daeafb884af6994bb4a7", size = 1115212, upload-time = "2026-09-24T18:34:39.945Z" }, + { url = "https://files.pythonhosted.org/packages/52/55/b78d9b56b7af04c8af8ab58e6d18c5977fd577b649b5caf698337af95b75/xgrammar-0.2.8-cp312-cp312-win_arm64.whl", hash = "sha256:2d4a9446b1b3e53fe4efcc5f40fafff0f79407c6b6fbd54a254909978808585e", size = 1084228, upload-time = "2026-09-24T18:34:41.521Z" }, + { url = "https://files.pythonhosted.org/packages/1b/06/b1fafc0cc26d74ff202a7c7900ba6a865679d3c873a2ca788a3f10f88c45/xgrammar-0.2.8-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:cb6fec998e8416f3079dcc29c39c5cfded0866ee2433016c4f1170f2fc8d440f", size = 1339887, upload-time = "2026-09-24T18:34:43.139Z" }, + { url = "https://files.pythonhosted.org/packages/67/c0/3690e8da12ae64b97ec1d69bf0802cc58661ccedf791e0b05e360083cafe/xgrammar-0.2.8-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:dec7222c239668cbe23fe84bf5faf035b9308983fed03907b6a1837fdad9e9f1", size = 1157369, upload-time = "2026-09-24T18:34:45.061Z" }, + { url = "https://files.pythonhosted.org/packages/71/be/bf1904b293fb3245347ea672b4c24d3320600fc3935f9e6bb840fa75681c/xgrammar-0.2.8-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5d24c79b88b3a6aba20d76d15119892f326f7fec85882946d6666b482cf45de4", size = 1311293, upload-time = "2026-09-24T18:34:46.741Z" }, + { url = "https://files.pythonhosted.org/packages/87/7a/d473f74ce72f8684857a986e8e60224e27acc96e89ff8599a30bd28b45a8/xgrammar-0.2.8-cp313-cp313-win_amd64.whl", hash = "sha256:08c80f20e18402397c11af6cee3c965da41f5116fa033115297bcbdaa877e07a", size = 1115212, upload-time = "2026-09-24T18:34:48.468Z" }, + { url = "https://files.pythonhosted.org/packages/eb/1e/613162d4cc6eda909d824b3591026151f93c4275ea7c836eb24dbd7bbd68/xgrammar-0.2.8-cp313-cp313-win_arm64.whl", hash = "sha256:ec85db49c8a48f2ee9dd8abb10eb09ee1f22d60b79348aefc5f23824e87791f0", size = 1084228, upload-time = "2026-09-24T18:34:50.205Z" }, + { url = "https://files.pythonhosted.org/packages/00/a0/580a9b6c18d3a790bc924917b43288b2e9b8a375ae27bf0c8a31494820bb/xgrammar-0.2.8-cp314-cp314-macosx_10_15_x86_64.whl", hash = "sha256:67ceb409708ba95479f833f58e6de963a95d29bf96d9114adcdf246e90afca06", size = 1496762, upload-time = "2026-09-24T18:34:51.957Z" }, + { url = "https://files.pythonhosted.org/packages/9a/65/76adc7a5436a996abde62d09d3148ea15be893618bb4eecbc9a76516c465/xgrammar-0.2.8-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:6236c482f9f34f1ce709cc67eff07363e0514b0bcc841210d4cc3b18b26e30f2", size = 1339888, upload-time = "2026-09-24T18:34:53.498Z" }, + { url = "https://files.pythonhosted.org/packages/3c/8d/5c59a22cc77d0bc4eadc65e3cf5c5365f0fee81881a84c73f649e12cf75c/xgrammar-0.2.8-cp314-cp314-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:45d242227001175bacd70031017ddef80e0c9c5141d11fd04bca5f08abee5417", size = 1157372, upload-time = "2026-09-24T18:34:55.578Z" }, + { url = "https://files.pythonhosted.org/packages/37/a1/8042e812c7dbd06cceff2aa8f61cf1c0d21fe51af7ca7407231c3101347f/xgrammar-0.2.8-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:dadcff0ecab59ffbc782e9db6d16f84f7b8df87bd2f28b9470136fcb4837f098", size = 1311291, upload-time = "2026-09-24T18:34:57.242Z" }, + { url = "https://files.pythonhosted.org/packages/58/a0/d13e351ef69a9df8e0808f475c17a2ecfc9e9e61f4b39a55395ec03105bf/xgrammar-0.2.8-cp314-cp314-win_amd64.whl", hash = "sha256:a9d917bfb075106f54533ee9e21c3791590c9277033ec0719d807a641987ddb0", size = 1142263, upload-time = "2026-09-24T18:34:58.887Z" }, + { url = "https://files.pythonhosted.org/packages/99/40/ec0405e6e8cd27f4da92a8b6bc7e5c26aaa9ec766dc5610db4bb00b8d936/xgrammar-0.2.8-cp314-cp314-win_arm64.whl", hash = "sha256:924626681e106933e90f9790b5d05b7c380acf17317d580d53c364cb4d19eee5", size = 1111958, upload-time = "2026-09-24T18:35:00.399Z" }, + { url = "https://files.pythonhosted.org/packages/c0/8e/5231ec249f11bdac46adf135ef70d9c113cce4fb0b6bcba31602777389e4/xgrammar-0.2.8-cp314-cp314t-macosx_10_15_x86_64.whl", hash = "sha256:4ae12599a1f8ddbf0ca0ab94b620f7e489de31f90c06b8396a9bd29adf22d784", size = 1496791, upload-time = "2026-09-24T18:35:01.933Z" }, + { url = "https://files.pythonhosted.org/packages/84/bd/985460d67505a709e39018ca713213d05b0fffc33e8ae12276ba35313463/xgrammar-0.2.8-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:6b2749c2ce1b73e7367569abd45c21aa359f5d6e7d8d88307b14cf5a8b716072", size = 1339847, upload-time = "2026-09-24T18:35:03.457Z" }, + { url = "https://files.pythonhosted.org/packages/d8/5f/6217b7553b21cea62381cf1fedb2dbf43010b7321f989ef496d401c5e338/xgrammar-0.2.8-cp314-cp314t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:47a041d9cb10f6c9da0d33db4aaa7b81163f677500278c4eab3e61b7ede5b2d3", size = 1157422, upload-time = "2026-09-24T18:35:05.168Z" }, + { url = "https://files.pythonhosted.org/packages/07/88/2d4feb6e4b347bec28d24890856b920176a15e5b4d6626a7986f8291c728/xgrammar-0.2.8-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:71bf243b0a7eec297857e8fd8abd0b7ea499968c3748568d5576ccf864ce29bc", size = 1311301, upload-time = "2026-09-24T18:35:06.666Z" }, + { url = "https://files.pythonhosted.org/packages/b1/3b/7ff3db56f4b8cddad15430a180931329f7bfe207a29410f9e16090838e18/xgrammar-0.2.8-cp314-cp314t-win_amd64.whl", hash = "sha256:9837e7948212fae1c12abc2b992305a9e333a266a1ae5150504738e1b9327c06", size = 1142263, upload-time = "2026-09-24T18:35:08.518Z" }, + { url = "https://files.pythonhosted.org/packages/84/5b/96e846071cbc77014896d9d3e91592a35afb1541dc8b89c5e393ddaa30e6/xgrammar-0.2.8-cp314-cp314t-win_arm64.whl", hash = "sha256:0c6125fa3dc27b46adde83329571c1dac2ab2474779daee207428ce2e9c6e9b5", size = 1111958, upload-time = "2026-09-24T18:35:10.366Z" }, +] + [[package]] name = "xmltodict" version = "1.0.4" @@ -10816,7 +12990,8 @@ name = "yarl" version = "1.25.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "idna" }, + { name = "idna", version = "3.15", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version < '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version < '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, + { name = "idna", version = "3.20", source = { registry = "https://pypi.org/simple" }, marker = "(python_full_version >= '3.11' and extra == 'extra-7-vec-inf-sglang') or (python_full_version >= '3.11' and extra == 'extra-7-vec-inf-vllm') or (extra == 'extra-7-vec-inf-sglang' and extra == 'extra-7-vec-inf-vllm')" }, { name = "multidict" }, { name = "propcache" }, ] diff --git a/vec_inf/cli/_helper.py b/vec_inf/cli/_helper.py index d1cbe8f2..65539bf2 100644 --- a/vec_inf/cli/_helper.py +++ b/vec_inf/cli/_helper.py @@ -328,7 +328,6 @@ class MetricsResponseFormatter: def __init__(self, metrics: Union[dict[str, float], str]): self.metrics = self._set_metrics(metrics) self.table = create_table("Metric", "Value") - self.enabled_prefix_caching = self._check_prefix_caching() def _set_metrics(self, metrics: Union[dict[str, float], str]) -> dict[str, float]: """Set the metrics attribute. @@ -345,16 +344,6 @@ def _set_metrics(self, metrics: Union[dict[str, float], str]) -> dict[str, float """ return metrics if isinstance(metrics, dict) else {} - def _check_prefix_caching(self) -> bool: - """Check if prefix caching is enabled. - - Returns - ------- - bool - True if prefix caching metrics are present - """ - return self.metrics.get("gpu_prefix_cache_hit_rate") is not None - def format_failed_metrics(self, message: str) -> None: """Format error message for failed metrics collection. @@ -371,8 +360,8 @@ def format_metrics(self) -> None: Formats and adds to the table: - Throughput metrics - Request queue metrics - - Cache usage metrics - - Prefix cache metrics (if enabled) + - KV cache usage and preemptions + - Prefix cache hit rate (if the server reports it) - Latency metrics - Token counts """ @@ -395,29 +384,21 @@ def format_metrics(self) -> None: "Requests Waiting", f"{self.metrics.get('requests_waiting', 0):.0f} reqs", ) - self.table.add_row( - "Requests Swapped", - f"{self.metrics.get('requests_swapped', 0):.0f} reqs", - ) - # Cache usage metrics + # KV cache metrics self.table.add_row( - "GPU Cache Usage", - f"{self.metrics.get('gpu_cache_usage', 0) * 100:.1f}%", + "KV Cache Usage", + f"{self.metrics.get('kv_cache_usage', 0) * 100:.1f}%", ) self.table.add_row( - "CPU Cache Usage", - f"{self.metrics.get('cpu_cache_usage', 0) * 100:.1f}%", + "Preemptions", + f"{self.metrics.get('preemptions_total', 0):.0f}", ) - if self.enabled_prefix_caching: - self.table.add_row( - "GPU Prefix Cache Hit Rate", - f"{self.metrics.get('gpu_prefix_cache_hit_rate', 0) * 100:.1f}%", - ) + if "prefix_cache_hit_rate" in self.metrics: self.table.add_row( - "CPU Prefix Cache Hit Rate", - f"{self.metrics.get('cpu_prefix_cache_hit_rate', 0) * 100:.1f}%", + "Prefix Cache Hit Rate", + f"{self.metrics['prefix_cache_hit_rate'] * 100:.1f}%", ) # Show average latency if available @@ -426,6 +407,11 @@ def format_metrics(self) -> None: "Avg Request Latency", f"{self.metrics['avg_request_latency']:.1f} s", ) + if "avg_queue_time" in self.metrics: + self.table.add_row( + "Avg Queue Time", + f"{self.metrics['avg_queue_time']:.2f} s", + ) # Token counts self.table.add_row( diff --git a/vec_inf/client/_client_vars.py b/vec_inf/client/_client_vars.py index 89dd327d..0f4e99b6 100644 --- a/vec_inf/client/_client_vars.py +++ b/vec_inf/client/_client_vars.py @@ -10,7 +10,9 @@ SRC_DIR : str Absolute path to the package source directory KEY_METRICS : dict - Mapping of vLLM metrics to their human-readable names + Mapping of vLLM and SGLang Prometheus metrics to output keys and aggregation +KEY_METRICS_EXCLUDED_LABELS : dict + Label values to skip when aggregating a metric SLURM_JOB_CONFIG_ARGS : dict Mapping of SLURM configuration arguments to their parameter names VLLM_SHORT_TO_LONG_MAP : dict @@ -24,19 +26,55 @@ SRC_DIR = str(Path(__file__).parent.parent) -# Key production metrics for inference servers -KEY_METRICS = { - "vllm:prompt_tokens_total": "total_prompt_tokens", - "vllm:generation_tokens_total": "total_generation_tokens", - "vllm:e2e_request_latency_seconds_sum": "request_latency_sum", - "vllm:e2e_request_latency_seconds_count": "request_latency_count", - "vllm:request_queue_time_seconds_sum": "queue_time_sum", - "vllm:request_success_total": "successful_requests_total", - "vllm:num_requests_running": "requests_running", - "vllm:num_requests_waiting": "requests_waiting", - "vllm:num_requests_swapped": "requests_swapped", - "vllm:gpu_cache_usage_perc": "gpu_cache_usage", - "vllm:cpu_cache_usage_perc": "cpu_cache_usage", +# Key production metrics for inference servers. +# +# Keys are Prometheus sample names as exposed on the server's /metrics endpoint. +# Values are (output key, aggregation). A metric is exported once per label set +# (per engine for vLLM data parallel, per finish reason for request counts, per +# rank for SGLang), so every matching sample is combined with the aggregation: +# "sum" adds them and "mean" averages them. +# +# vLLM names match vllm/v1/metrics/loggers.py in vLLM 0.32.0. vLLM V1 no longer +# exports gpu_cache_usage_perc, cpu_cache_usage_perc, num_requests_swapped or +# the *_prefix_cache_hit_rate gauges; the prefix cache hit rate is derived from +# the prefix_cache_hits / prefix_cache_queries counters instead. +# SGLang names match sglang/srt/observability/metrics_collector.py in SGLang +# 0.5.12.post1. SGLang only serves /metrics when launched with --enable-metrics. +KEY_METRICS: dict[str, tuple[str, str]] = { + # vLLM + "vllm:prompt_tokens_total": ("total_prompt_tokens", "sum"), + "vllm:generation_tokens_total": ("total_generation_tokens", "sum"), + "vllm:e2e_request_latency_seconds_sum": ("request_latency_sum", "sum"), + "vllm:e2e_request_latency_seconds_count": ("request_latency_count", "sum"), + "vllm:request_queue_time_seconds_sum": ("queue_time_sum", "sum"), + "vllm:request_queue_time_seconds_count": ("queue_time_count", "sum"), + "vllm:request_success_total": ("successful_requests_total", "sum"), + "vllm:num_requests_running": ("requests_running", "sum"), + "vllm:num_requests_waiting": ("requests_waiting", "sum"), + "vllm:num_preemptions_total": ("preemptions_total", "sum"), + "vllm:kv_cache_usage_perc": ("kv_cache_usage", "mean"), + "vllm:prefix_cache_queries_total": ("prefix_cache_queries", "sum"), + "vllm:prefix_cache_hits_total": ("prefix_cache_hits", "sum"), + # SGLang + "sglang:prompt_tokens_total": ("total_prompt_tokens", "sum"), + "sglang:generation_tokens_total": ("total_generation_tokens", "sum"), + "sglang:e2e_request_latency_seconds_sum": ("request_latency_sum", "sum"), + "sglang:e2e_request_latency_seconds_count": ("request_latency_count", "sum"), + "sglang:queue_time_seconds_sum": ("queue_time_sum", "sum"), + "sglang:queue_time_seconds_count": ("queue_time_count", "sum"), + "sglang:num_requests_total": ("successful_requests_total", "sum"), + "sglang:num_running_reqs": ("requests_running", "sum"), + "sglang:num_queue_reqs": ("requests_waiting", "sum"), + "sglang:num_retracted_requests_total": ("preemptions_total", "sum"), + "sglang:token_usage": ("kv_cache_usage", "mean"), + "sglang:cache_hit_rate": ("prefix_cache_hit_rate", "mean"), +} + +# Samples to skip when aggregating, as {metric name: {label: excluded values}}. +# vllm:request_success_total is labelled by finished_reason, and requests that +# were aborted by the client or hit an internal error are not successes. +KEY_METRICS_EXCLUDED_LABELS: dict[str, dict[str, set[str]]] = { + "vllm:request_success_total": {"finished_reason": {"abort", "error"}}, } # Slurm job configuration arguments @@ -57,7 +95,7 @@ "error": "err_file", } -# vLLM engine args mapping between short and long names +# vLLM engine args mapping between short and long names (vLLM 0.32.0 aliases) VLLM_SHORT_TO_LONG_MAP = { "-tp": "--tensor-parallel-size", "-pp": "--pipeline-parallel-size", @@ -74,11 +112,20 @@ "-dpb": "--data-parallel-backend", "-dph": "--data-parallel-hybrid-lb", "-dpe": "--data-parallel-external-lb", - "-O": "--compilation-config", + "-dpm": "--data-parallel-multi-port-external-lb", + "-asc": "--api-server-count", + "-ep": "--enable-expert-parallel", + "-sc": "--speculative-config", + "-cc": "--compilation-config", + "-ac": "--attention-config", + # vLLM expands -O (e.g. -O3) to --optimization-level + "-O": "--optimization-level", "-q": "--quantization", } -# SGLang engine args mapping between short and long names +# SGLang engine args mapping between short and long names. SGLang itself accepts +# --tp-size, --pp-size, --dp-size, --ep-size and --ep as aliases; --tp, --pp and +# --dp are vec-inf shorthands. SGLANG_SHORT_TO_LONG_MAP = { "--tp": "--tensor-parallel-size", "--tp-size": "--tensor-parallel-size", @@ -87,7 +134,7 @@ "--dp": "--data-parallel-size", "--dp-size": "--data-parallel-size", "--ep": "--expert-parallel-size", - "--ep-size": "--expert-parallel-expert-size", + "--ep-size": "--expert-parallel-size", } # Mapping of engine short names to their argument mappings diff --git a/vec_inf/client/_helper.py b/vec_inf/client/_helper.py index dbb64a20..d85596e9 100644 --- a/vec_inf/client/_helper.py +++ b/vec_inf/client/_helper.py @@ -5,6 +5,7 @@ """ import json +import re import time import warnings from pathlib import Path @@ -19,6 +20,7 @@ BATCH_MODE_REQUIRED_MATCHING_ARGS, ENGINE_SHORT_TO_LONG_MAP, KEY_METRICS, + KEY_METRICS_EXCLUDED_LABELS, SRC_DIR, SUPPORTED_ENGINES, ) @@ -44,6 +46,14 @@ ) +# One Prometheus sample line: name, optional {labels}, value (timestamp ignored) +_PROM_SAMPLE_RE = re.compile( + r"^([a-zA-Z_:][a-zA-Z0-9_:]*)(?:\{(.*)\})?\s+(\S+)(?:\s+\S+)?$" +) +# One label pair inside the braces, allowing escaped quotes in the value +_PROM_LABEL_RE = re.compile(r'([a-zA-Z_][a-zA-Z0-9_]*)="((?:[^"\\]|\\.)*)"') + + class ModelLauncher: """Helper class for handling inference server launch. @@ -167,19 +177,24 @@ def _process_engine_args( engine_args: dict[str, str | bool] = {} engine_arg_map = ENGINE_SHORT_TO_LONG_MAP[engine_choice] - for arg in arg_string.split(","): - if "=" in arg: - key, value = arg.split("=") - if key.strip() in engine_arg_map: - key = engine_arg_map[key.strip()] - engine_args[key.strip()] = value.strip() - elif "-O" in arg.strip(): + for raw_arg in arg_string.split(","): + arg = raw_arg.strip() + if not arg: + continue + if arg.startswith("-O"): + # vLLM optimization level shorthand: -O3, -O=3 or -O 3 if engine_choice != "vllm": raise ValueError("-O is only supported for vLLM") - key = engine_arg_map["-O"] - engine_args[key] = arg.strip()[2:].strip() + level = arg[2:].lstrip("= ").strip() + if not level: + raise ValueError("-O requires an optimization level, e.g. -O3") + engine_args[engine_arg_map["-O"]] = level + elif "=" in arg: + key, value = arg.split("=", 1) + key = key.strip() + engine_args[engine_arg_map.get(key, key)] = value.strip() else: - engine_args[arg.strip()] = True + engine_args[engine_arg_map.get(arg, arg)] = True return engine_args def _process_env_vars(self, env_arg: str) -> dict[str, str]: @@ -949,7 +964,6 @@ def __init__(self, slurm_job_id: str): self.status_info = self._get_status_info() self.log_dir = self.status_info.log_dir self.metrics_url = self._build_metrics_url() - self.enabled_prefix_caching = self._check_prefix_caching() self._prev_prompt_tokens: float = 0.0 self._prev_generation_tokens: float = 0.0 @@ -992,26 +1006,32 @@ def _build_metrics_url(self) -> str: (parsed.scheme, parsed.netloc, f"{clean_path}/metrics", "", "", "") ) - def _check_prefix_caching(self) -> bool: - """Check if prefix caching is enabled. + @staticmethod + def _parse_labels(label_text: str) -> dict[str, str]: + """Parse the label set of a Prometheus sample. + + Parameters + ---------- + label_text : str + Text between the braces, e.g. ``engine="0",model_name="m"`` Returns ------- - bool - True if prefix caching is enabled, False otherwise + dict[str, str] + Label names mapped to their (unescaped) values """ - job_json = utils.read_slurm_log( - self.status_info.model_name, - self.slurm_job_id, - "json", - self.log_dir, - ) - if isinstance(job_json, str): - return False - return bool(cast(dict[str, str], job_json).get("enable_prefix_caching", False)) + return { + name: value.replace('\\"', '"').replace("\\\\", "\\") + for name, value in _PROM_LABEL_RE.findall(label_text) + } def _parse_metrics(self, metrics_text: str) -> dict[str, float]: - """Parse metrics with latency count and sum. + """Parse Prometheus text exposition into vec-inf metric values. + + Every sample of a tracked metric is aggregated across its label sets + (engines, finish reasons, ranks) as configured in ``KEY_METRICS``. The + prefix cache hit rate is derived from the hit and query counters when + the server does not export it directly. Parameters ---------- @@ -1023,27 +1043,42 @@ def _parse_metrics(self, metrics_text: str) -> dict[str, float]: dict[str, float] Parsed metrics as key-value pairs """ - key_metrics = KEY_METRICS - - if self.enabled_prefix_caching: - key_metrics["vllm:gpu_prefix_cache_hit_rate"] = "gpu_prefix_cache_hit_rate" - key_metrics["vllm:cpu_prefix_cache_hit_rate"] = "cpu_prefix_cache_hit_rate" - - parsed: dict[str, float] = {} - for line in metrics_text.split("\n"): - if line.startswith("#") or not line.strip(): + samples: dict[str, list[float]] = {} + for raw_line in metrics_text.splitlines(): + line = raw_line.strip() + if not line or line.startswith("#"): continue - parts = line.split() - if len(parts) < 2: + match = _PROM_SAMPLE_RE.match(line) + if not match: + continue + metric_name, label_text, value_text = match.groups() + if metric_name not in KEY_METRICS: continue - metric_name = parts[0].split("{")[0] - if metric_name in key_metrics: - try: - parsed[key_metrics[metric_name]] = float(parts[1]) - except (ValueError, IndexError): + excluded = KEY_METRICS_EXCLUDED_LABELS.get(metric_name) + if excluded and label_text: + labels = self._parse_labels(label_text) + if any(labels.get(k) in values for k, values in excluded.items()): continue + + try: + value = float(value_text) + except ValueError: + continue + samples.setdefault(metric_name, []).append(value) + + parsed: dict[str, float] = {} + for metric_name, values in samples.items(): + key, aggregation = KEY_METRICS[metric_name] + total = sum(values) + parsed[key] = total / len(values) if aggregation == "mean" else total + + if "prefix_cache_hit_rate" not in parsed and "prefix_cache_queries" in parsed: + queries = parsed["prefix_cache_queries"] + parsed["prefix_cache_hit_rate"] = ( + parsed.get("prefix_cache_hits", 0.0) / queries if queries > 0 else 0.0 + ) return parsed def fetch_metrics(self) -> Union[dict[str, float], str]: @@ -1121,6 +1156,18 @@ def fetch_metrics(self) -> Union[dict[str, float], str]: latency_sum / latency_count if latency_count > 0 else 0.0 ) + # Calculate average queue time if data is available + if ( + "queue_time_sum" in current_metrics + and "queue_time_count" in current_metrics + ): + queue_count = current_metrics["queue_time_count"] + current_metrics["avg_queue_time"] = ( + current_metrics["queue_time_sum"] / queue_count + if queue_count > 0 + else 0.0 + ) + return current_metrics except requests.RequestException as e: diff --git a/vec_inf/client/_slurm_script_generator.py b/vec_inf/client/_slurm_script_generator.py index dcca77d8..7b9a3952 100644 --- a/vec_inf/client/_slurm_script_generator.py +++ b/vec_inf/client/_slurm_script_generator.py @@ -197,8 +197,9 @@ def _generate_launch_cmd(self) -> str: else: launch_cmd.append(f" {arg} {value} \\") - # A known bug in vLLM requires setting backend to ray for multi-node - # Remove this when the bug is fixed + # Multi-node vLLM runs on the Ray cluster started in the server setup + # section, so the Ray executor must be selected explicitly. vLLM only + # picks Ray on its own when it is already connected to a Ray cluster. if self.is_multinode: launch_cmd.append(" --distributed-executor-backend ray \\") diff --git a/vec_inf/client/_slurm_templates.py b/vec_inf/client/_slurm_templates.py index 1a7bf496..e9537c51 100644 --- a/vec_inf/client/_slurm_templates.py +++ b/vec_inf/client/_slurm_templates.py @@ -209,6 +209,8 @@ class SlurmScriptTemplate(TypedDict): " --served-model-name {model_name} \\", ' --host "0.0.0.0" \\', " --port $server_port_number \\", + # SGLang only serves /metrics (used by `vec-inf metrics`) with this flag + " --enable-metrics \\", ], "sglang_multinode": [ "for ((i = 0; i < $SLURM_JOB_NUM_NODES; i++)); do", @@ -221,6 +223,7 @@ class SlurmScriptTemplate(TypedDict): " --served-model-name {model_name} \\", ' --host "0.0.0.0" \\', " --port $server_port_number \\", + " --enable-metrics \\", ' --nccl-init-addr "$NCCL_INIT_ADDR" \\', " --nnodes {num_nodes} \\", ' --node-rank "$i" \\', @@ -332,6 +335,8 @@ class BatchModelLaunchScriptTemplate(TypedDict): " --served-model-name {model_name} \\", ' --host "0.0.0.0" \\', " --port $server_port_number \\", + # SGLang only serves /metrics (used by `vec-inf metrics`) with this flag + " --enable-metrics \\", ], }, } diff --git a/vec_inf/client/_utils.py b/vec_inf/client/_utils.py index e722f12a..4c0ed5c7 100644 --- a/vec_inf/client/_utils.py +++ b/vec_inf/client/_utils.py @@ -6,6 +6,7 @@ import json import os +import re import subprocess import warnings from pathlib import Path @@ -21,6 +22,19 @@ from vec_inf.client.models import ModelStatus +# Log lines that carry an explicit non-error level are never treated as fatal, +# even when they mention a failure. For example Ray logs a warning that it +# failed to connect to its observability pubsub during multi-node startup. +# Patterns, in order: vLLM log lines (level then MM-DD HH:MM:SS), Ray Python +# logging (date, time, then level) and Ray C++ glog lines (single letter level +# inside the leading bracket). +_NON_ERROR_LOG_LINE_PATTERNS = ( + re.compile(r"\b(?:DEBUG|INFO|WARNING) \d{2}-\d{2} \d{2}:\d{2}:\d{2}\b"), + re.compile(r"^\d{4}-\d{2}-\d{2} [\d:,.]+\s+(?:DEBUG|INFO|WARNING)\s"), + re.compile(r"^\[\d{4}-\d{2}-\d{2} [\d:,.]+ [DIW] \d+ \d+\]"), +) + + def run_bash_command(command: str) -> tuple[str, str]: """Run a bash command and return the output. @@ -145,6 +159,9 @@ def is_server_running( # Skip lines that match ignore patterns if any(ignore_pattern in line_lower for ignore_pattern in ignore_patterns): continue + # Skip log lines with an explicit non-error level + if any(p.search(line) for p in _NON_ERROR_LOG_LINE_PATTERNS): + continue # Check for fatal error patterns for pattern in fatal_error_patterns: diff --git a/vllm.Dockerfile b/vllm.Dockerfile index 3727593b..7f60c16d 100644 --- a/vllm.Dockerfile +++ b/vllm.Dockerfile @@ -1,4 +1,8 @@ -FROM nvidia/cuda:12.8.1-cudnn-devel-ubuntu24.04 +# vLLM 0.32 ships torch 2.13 built for CUDA 13.0, and FlashInfer compiles some +# kernels at runtime with the system nvcc. Keep the toolkit on the same CUDA +# major version as torch (this matches vLLM's own image). CUDA 13 needs NVIDIA +# driver 580 or newer on the GPU nodes. +FROM nvidia/cuda:13.0.3-cudnn-devel-ubuntu24.04 # Non-interactive apt-get commands ARG DEBIAN_FRONTEND=noninteractive @@ -80,7 +84,8 @@ RUN uv export --frozen --no-emit-project --no-hashes \ # ABI mismatch in :0.19.0 at build time instead of at job start. RUN python3.12 -c "import vllm, datasets, pyarrow, transformers, torch; \ print('vllm', vllm.__version__, '/ datasets', datasets.__version__, \ - '/ pyarrow', pyarrow.__version__, '/ torch', torch.__version__)" + '/ pyarrow', pyarrow.__version__, '/ torch', torch.__version__, \ + '/ torch CUDA', torch.version.cuda)" # Install a single, system NCCL (from NVIDIA CUDA repo in base image) RUN apt-get update && apt-get install -y --allow-change-held-packages\ From af91591b994a2c3aff1c38432f290e94e4db5a21 Mon Sep 17 00:00:00 2001 From: Amrit Krishnan Date: Sun, 25 Oct 2026 13:15:47 -0400 Subject: [PATCH 02/10] Anchor the backend version lookup in the docker workflow The vllm extra now carries a python_full_version marker, so the unanchored grep also matched dependency entries and wrote several lines to GITHUB_OUTPUT, failing the step. --- .github/workflows/docker.yml | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/.github/workflows/docker.yml b/.github/workflows/docker.yml index af8287dc..ed6c9928 100644 --- a/.github/workflows/docker.yml +++ b/.github/workflows/docker.yml @@ -59,7 +59,11 @@ jobs: - name: Extract backend version id: backend-version run: | - VERSION=$(grep -A 1 "name = \"${{ matrix.backend }}\"" uv.lock | grep version | cut -d '"' -f 2) + # Anchor on the package header line. Unanchored, the pattern also + # matches dependency entries such as `{ name = "vllm", marker = + # "python_full_version >= '3.11' ..." }`, whose marker contains + # "version", and the step writes several lines to GITHUB_OUTPUT. + VERSION=$(grep -A 1 "^name = \"${{ matrix.backend }}\"$" uv.lock | grep '^version = ' | head -n 1 | cut -d '"' -f 2) echo "version=$VERSION" >> $GITHUB_OUTPUT - name: Compute image base path From f6a0315c1a82e7415415fe0988866bc2a426e637 Mon Sep 17 00:00:00 2001 From: Amrit Krishnan Date: Sun, 25 Oct 2026 13:32:32 -0400 Subject: [PATCH 03/10] Test metrics parsing against live vllm 0.32 output, fix sglang kernel pin, docs - Add a /metrics capture from vllm 0.32.0 serving Llama-3.2-1B-Instruct on two A40 nodes (pipeline parallel 2) as a test fixture, plus tests for data parallel aggregation, finish reasons, label escaping and SGLang metric names. - Test that generated scripts select the Ray executor only for multi-node vllm, that SGLang always gets --enable-metrics, and that the generated scripts are valid bash. - Drop sgl_kernel from the sglang extra. sglang pins its own sglang-kernel package, and the older sgl-kernel distribution installs the same sgl_kernel module built for a different torch, so importing it failed with an undefined c10 symbol. - Update the metrics section, engine argument aliases and the vllm docs link in the user guide. --- docs/user_guide.md | 46 +- pyproject.toml | 5 +- .../client/fixtures/vllm_0.32.0_metrics.txt | 654 ++++++++++++++++++ tests/vec_inf/client/test_helper.py | 144 ++++ .../client/test_slurm_script_generator.py | 46 ++ uv.lock | 11 - 6 files changed, 876 insertions(+), 30 deletions(-) create mode 100644 tests/vec_inf/client/fixtures/vllm_0.32.0_metrics.txt diff --git a/docs/user_guide.md b/docs/user_guide.md index 79e8dea7..3d4b3cd3 100644 --- a/docs/user_guide.md +++ b/docs/user_guide.md @@ -76,9 +76,11 @@ vec-inf launch Qwen2.5-3B-Instruct \ For the full list of inference engine arguments, you can find them here: -* [vLLM: `vllm serve` Arguments](https://docs.vllm.ai/en/stable/serving/engine_args.html) +* [vLLM: `vllm serve` Arguments](https://docs.vllm.ai/en/v0.32.0/configuration/engine_args.html) * [SGLang: Server Arguments](https://docs.sglang.io/advanced_features/server_arguments.html) +Short aliases are expanded to the long argument names before launch. For vLLM these are the aliases `vllm serve` itself accepts, for example `-tp`, `-pp`, `-dp`, `-ep` (`--enable-expert-parallel`), `-cc` (`--compilation-config`), `-q` (`--quantization`) and `-O3` (`--optimization-level 3`). For SGLang, `--tp`, `--pp`, `--dp` and `--ep`/`--ep-size` map to `--tensor-parallel-size`, `--pipeline-parallel-size`, `--data-parallel-size` and `--expert-parallel-size`. + #### Custom models You can also launch your own custom model as long as the model architecture is supported by the underlying inference engine, and make sure to follow the instructions below: @@ -241,23 +243,32 @@ vec-inf metrics 15373800 And you will see the performance metrics streamed to your console, note that the metrics are updated with a 2-second interval. ``` -┏━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓ -┃ Metric ┃ Value ┃ -┡━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩ -│ Prompt Throughput │ 10.9 tokens/s │ -│ Generation Throughput │ 34.2 tokens/s │ -│ Requests Running │ 1 reqs │ -│ Requests Waiting │ 0 reqs │ -│ Requests Swapped │ 0 reqs │ -│ GPU Cache Usage │ 0.1% │ -│ CPU Cache Usage │ 0.0% │ -│ Avg Request Latency │ 2.6 s │ -│ Total Prompt Tokens │ 441 tokens │ -│ Total Generation Tokens │ 1748 tokens │ -│ Successful Requests │ 14 reqs │ -└─────────────────────────┴─────────────────┘ +┏━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ +┃ Metric ┃ Value ┃ +┡━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ +│ Prompt Throughput │ 16.9 tokens/s │ +│ Generation Throughput │ 50.8 tokens/s │ +│ Requests Running │ 0 reqs │ +│ Requests Waiting │ 0 reqs │ +│ KV Cache Usage │ 0.0% │ +│ Preemptions │ 0 │ +│ Prefix Cache Hit Rate │ 0.0% │ +│ Avg Request Latency │ 0.1 s │ +│ Avg Queue Time │ 0.00 s │ +│ Total Prompt Tokens │ 69 tokens │ +│ Total Generation Tokens │ 208 tokens │ +│ Successful Requests │ 8 reqs │ +└─────────────────────────┴───────────────┘ ``` +**Notes** + +* The metrics come from the server's Prometheus `/metrics` endpoint. Counters are summed over all engines (for example with `--data-parallel-size`), and KV cache usage is averaged over them. +* KV cache usage is the share of the GPU KV cache in use. vLLM 0.32 no longer reports separate GPU and CPU cache usage or swapped requests, so those rows were replaced by **KV Cache Usage** and **Preemptions**. +* The prefix cache hit rate is the share of prompt tokens served from the prefix cache since the server started. Prefix caching is on by default in vLLM and SGLang, so the row is shown whenever the server reports it. +* **Successful Requests** counts requests that finished normally. Requests aborted by the client or failed with an error are not counted. +* SGLang servers launched by `vec-inf` are started with `--enable-metrics`, which SGLang needs before it serves `/metrics`. + ### `shutdown` command Finally, when you're finished using a model, you can shut it down by providing the Slurm job ID: @@ -347,8 +358,7 @@ Once the inference server is ready, you can start sending in inference requests. "content":"Arrr, I be Captain Chatbeard, the scurviest chatbot on the seven seas! Ye be wantin' to know me identity, eh? Well, matey, I be a swashbucklin' AI, here to provide ye with answers and swappin' tales, savvy?", "role":"assistant", "function_call":null, - "tool_calls":[], - "reasoning_content":null + "reasoning":null }, "stop_reason":null } diff --git a/pyproject.toml b/pyproject.toml index 547c4a26..b4032ae5 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -61,8 +61,11 @@ vllm = [ # SGLang inference backend (conflicts with vllm due to dependency version conflicts) # Install with: uv sync --extra sglang --group inference sglang = [ + # sglang pins its own kernel package (sglang-kernel). Do not add the older + # sgl-kernel distribution here: both install the sgl_kernel module, and the + # old one is built for a different torch, so importing it fails with an + # undefined c10 symbol. "sglang>=0.5.5", - "sgl_kernel>=0.3.0", "orjson>=3.11.0", "torchao>=0.9.0", ] diff --git a/tests/vec_inf/client/fixtures/vllm_0.32.0_metrics.txt b/tests/vec_inf/client/fixtures/vllm_0.32.0_metrics.txt new file mode 100644 index 00000000..020a966d --- /dev/null +++ b/tests/vec_inf/client/fixtures/vllm_0.32.0_metrics.txt @@ -0,0 +1,654 @@ +# HELP python_gc_objects_collected_total Objects collected during gc +# TYPE python_gc_objects_collected_total counter +python_gc_objects_collected_total{generation="0"} 16063.0 +python_gc_objects_collected_total{generation="1"} 3103.0 +python_gc_objects_collected_total{generation="2"} 421.0 +# HELP python_gc_objects_uncollectable_total Uncollectable objects found during GC +# TYPE python_gc_objects_uncollectable_total counter +python_gc_objects_uncollectable_total{generation="0"} 0.0 +python_gc_objects_uncollectable_total{generation="1"} 0.0 +python_gc_objects_uncollectable_total{generation="2"} 0.0 +# HELP python_gc_collections_total Number of times this generation was collected +# TYPE python_gc_collections_total counter +python_gc_collections_total{generation="0"} 1837.0 +python_gc_collections_total{generation="1"} 167.0 +python_gc_collections_total{generation="2"} 10.0 +# HELP python_info Python platform information +# TYPE python_info gauge +python_info{implementation="CPython",major="3",minor="12",patchlevel="3",version="3.12.3"} 1.0 +# HELP process_virtual_memory_bytes Virtual memory size in bytes. +# TYPE process_virtual_memory_bytes gauge +process_virtual_memory_bytes 8.210604032e+09 +# HELP process_resident_memory_bytes Resident memory size in bytes. +# TYPE process_resident_memory_bytes gauge +process_resident_memory_bytes 1.602420736e+09 +# HELP process_start_time_seconds Start time of the process since unix epoch in seconds. +# TYPE process_start_time_seconds gauge +process_start_time_seconds 1.79294880962e+09 +# HELP process_cpu_seconds_total Total user and system CPU time spent in seconds. +# TYPE process_cpu_seconds_total counter +process_cpu_seconds_total 40.59 +# HELP process_open_fds Number of open file descriptors. +# TYPE process_open_fds gauge +process_open_fds 93.0 +# HELP process_max_fds Maximum number of open file descriptors. +# TYPE process_max_fds gauge +process_max_fds 131072.0 +# HELP vllm:estimated_flops_per_gpu_total Estimated number of floating point operations per GPU (for Model Flops Utilization calculations). +# TYPE vllm:estimated_flops_per_gpu_total counter +vllm:estimated_flops_per_gpu_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.0 +# HELP vllm:estimated_flops_per_gpu_created Estimated number of floating point operations per GPU (for Model Flops Utilization calculations). +# TYPE vllm:estimated_flops_per_gpu_created gauge +vllm:estimated_flops_per_gpu_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428893409e+09 +# HELP vllm:estimated_read_bytes_per_gpu_total Estimated number of bytes read from memory per GPU (for Model Flops Utilization calculations). +# TYPE vllm:estimated_read_bytes_per_gpu_total counter +vllm:estimated_read_bytes_per_gpu_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.0 +# HELP vllm:estimated_read_bytes_per_gpu_created Estimated number of bytes read from memory per GPU (for Model Flops Utilization calculations). +# TYPE vllm:estimated_read_bytes_per_gpu_created gauge +vllm:estimated_read_bytes_per_gpu_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428893938e+09 +# HELP vllm:estimated_write_bytes_per_gpu_total Estimated number of bytes written to memory per GPU (for Model Flops Utilization calculations). +# TYPE vllm:estimated_write_bytes_per_gpu_total counter +vllm:estimated_write_bytes_per_gpu_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.0 +# HELP vllm:estimated_write_bytes_per_gpu_created Estimated number of bytes written to memory per GPU (for Model Flops Utilization calculations). +# TYPE vllm:estimated_write_bytes_per_gpu_created gauge +vllm:estimated_write_bytes_per_gpu_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.792949042889438e+09 +# HELP vllm:num_requests_running Number of requests in model execution batches. +# TYPE vllm:num_requests_running gauge +vllm:num_requests_running{engine="0",model_name="Llama-3.2-1B-Instruct"} 5.0 +# HELP vllm:num_requests_waiting Number of requests waiting to be processed. +# TYPE vllm:num_requests_waiting gauge +vllm:num_requests_waiting{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.0 +# HELP vllm:num_requests_waiting_by_reason Number of waiting requests by reason. Reason labels: 'capacity' = waiting for scheduling capacity; 'deferred' = deferred by transient constraints (LoRA budget, KV transfer, blocked status). Sum of all reasons equals vllm:num_requests_waiting. +# TYPE vllm:num_requests_waiting_by_reason gauge +vllm:num_requests_waiting_by_reason{engine="0",model_name="Llama-3.2-1B-Instruct",reason="capacity"} 0.0 +vllm:num_requests_waiting_by_reason{engine="0",model_name="Llama-3.2-1B-Instruct",reason="deferred"} 0.0 +# HELP vllm:engine_sleep_state Engine sleep state; awake = 0 means engine is sleeping; awake = 1 means engine is awake; weights_offloaded = 1 means sleep level 1; discard_all = 1 means sleep level 2. +# TYPE vllm:engine_sleep_state gauge +vllm:engine_sleep_state{engine="0",model_name="Llama-3.2-1B-Instruct",sleep_state="awake"} 1.0 +vllm:engine_sleep_state{engine="0",model_name="Llama-3.2-1B-Instruct",sleep_state="weights_offloaded"} 0.0 +vllm:engine_sleep_state{engine="0",model_name="Llama-3.2-1B-Instruct",sleep_state="discard_all"} 0.0 +# HELP vllm:kv_cache_usage_perc KV-cache usage. 1 means 100 percent usage. +# TYPE vllm:kv_cache_usage_perc gauge +vllm:kv_cache_usage_perc{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.0001261901306698343 +# HELP vllm:prefix_cache_queries_total Prefix cache queries, in terms of number of queried tokens. +# TYPE vllm:prefix_cache_queries_total counter +vllm:prefix_cache_queries_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 1307.0 +# HELP vllm:prefix_cache_queries_created Prefix cache queries, in terms of number of queried tokens. +# TYPE vllm:prefix_cache_queries_created gauge +vllm:prefix_cache_queries_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.792949042889909e+09 +# HELP vllm:prefix_cache_hits_total Prefix cache hits, in terms of number of cached tokens. +# TYPE vllm:prefix_cache_hits_total counter +vllm:prefix_cache_hits_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 960.0 +# HELP vllm:prefix_cache_hits_created Prefix cache hits, in terms of number of cached tokens. +# TYPE vllm:prefix_cache_hits_created gauge +vllm:prefix_cache_hits_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428899472e+09 +# HELP vllm:external_prefix_cache_queries_total External prefix cache queries from KV connector cross-instance cache sharing, in terms of number of queried tokens. +# TYPE vllm:external_prefix_cache_queries_total counter +vllm:external_prefix_cache_queries_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.0 +# HELP vllm:external_prefix_cache_queries_created External prefix cache queries from KV connector cross-instance cache sharing, in terms of number of queried tokens. +# TYPE vllm:external_prefix_cache_queries_created gauge +vllm:external_prefix_cache_queries_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428899899e+09 +# HELP vllm:external_prefix_cache_hits_total External prefix cache hits from KV connector cross-instance cache sharing, in terms of number of cached tokens. +# TYPE vllm:external_prefix_cache_hits_total counter +vllm:external_prefix_cache_hits_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.0 +# HELP vllm:external_prefix_cache_hits_created External prefix cache hits from KV connector cross-instance cache sharing, in terms of number of cached tokens. +# TYPE vllm:external_prefix_cache_hits_created gauge +vllm:external_prefix_cache_hits_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.792949042890029e+09 +# HELP vllm:mm_cache_queries_total Multi-modal cache queries, in terms of number of queried items. +# TYPE vllm:mm_cache_queries_total counter +vllm:mm_cache_queries_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.0 +# HELP vllm:mm_cache_queries_created Multi-modal cache queries, in terms of number of queried items. +# TYPE vllm:mm_cache_queries_created gauge +vllm:mm_cache_queries_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.792949042890066e+09 +# HELP vllm:mm_cache_hits_total Multi-modal cache hits, in terms of number of cached items. +# TYPE vllm:mm_cache_hits_total counter +vllm:mm_cache_hits_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.0 +# HELP vllm:mm_cache_hits_created Multi-modal cache hits, in terms of number of cached items. +# TYPE vllm:mm_cache_hits_created gauge +vllm:mm_cache_hits_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428901024e+09 +# HELP vllm:num_preemptions_total Cumulative number of preemption from the engine. +# TYPE vllm:num_preemptions_total counter +vllm:num_preemptions_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.0 +# HELP vllm:num_preemptions_created Cumulative number of preemption from the engine. +# TYPE vllm:num_preemptions_created gauge +vllm:num_preemptions_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428901386e+09 +# HELP vllm:prompt_tokens_total Number of prefill tokens processed. +# TYPE vllm:prompt_tokens_total counter +vllm:prompt_tokens_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 572.0 +# HELP vllm:prompt_tokens_created Number of prefill tokens processed. +# TYPE vllm:prompt_tokens_created gauge +vllm:prompt_tokens_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428904426e+09 +# HELP vllm:prompt_tokens_by_source_total Number of prompt tokens by source. +# TYPE vllm:prompt_tokens_by_source_total counter +vllm:prompt_tokens_by_source_total{engine="0",model_name="Llama-3.2-1B-Instruct",source="local_compute"} 332.0 +vllm:prompt_tokens_by_source_total{engine="0",model_name="Llama-3.2-1B-Instruct",source="local_cache_hit"} 240.0 +vllm:prompt_tokens_by_source_total{engine="0",model_name="Llama-3.2-1B-Instruct",source="external_kv_transfer"} 0.0 +# HELP vllm:prompt_tokens_by_source_created Number of prompt tokens by source. +# TYPE vllm:prompt_tokens_by_source_created gauge +vllm:prompt_tokens_by_source_created{engine="0",model_name="Llama-3.2-1B-Instruct",source="local_compute"} 1.7929490428904867e+09 +vllm:prompt_tokens_by_source_created{engine="0",model_name="Llama-3.2-1B-Instruct",source="local_cache_hit"} 1.792949042890522e+09 +vllm:prompt_tokens_by_source_created{engine="0",model_name="Llama-3.2-1B-Instruct",source="external_kv_transfer"} 1.7929490428905413e+09 +# HELP vllm:prompt_tokens_cached_total Number of cached prompt tokens (local + external). +# TYPE vllm:prompt_tokens_cached_total counter +vllm:prompt_tokens_cached_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 240.0 +# HELP vllm:prompt_tokens_cached_created Number of cached prompt tokens (local + external). +# TYPE vllm:prompt_tokens_cached_created gauge +vllm:prompt_tokens_cached_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428905787e+09 +# HELP vllm:prompt_tokens_cached_by_source_total Prefix-cache hit tokens by the cache tier that supplied them: device (local HBM), host (offloaded to DRAM), disk, p2p (transferred from another vLLM instance) or external_unspecified. Counted at admission; sums to vllm:prefix_cache_hits + vllm:external_prefix_cache_hits. +# TYPE vllm:prompt_tokens_cached_by_source_total counter +vllm:prompt_tokens_cached_by_source_total{engine="0",model_name="Llama-3.2-1B-Instruct",source="device"} 960.0 +vllm:prompt_tokens_cached_by_source_total{engine="0",model_name="Llama-3.2-1B-Instruct",source="host"} 0.0 +vllm:prompt_tokens_cached_by_source_total{engine="0",model_name="Llama-3.2-1B-Instruct",source="p2p"} 0.0 +vllm:prompt_tokens_cached_by_source_total{engine="0",model_name="Llama-3.2-1B-Instruct",source="disk"} 0.0 +vllm:prompt_tokens_cached_by_source_total{engine="0",model_name="Llama-3.2-1B-Instruct",source="external_unspecified"} 0.0 +# HELP vllm:prompt_tokens_cached_by_source_created Prefix-cache hit tokens by the cache tier that supplied them: device (local HBM), host (offloaded to DRAM), disk, p2p (transferred from another vLLM instance) or external_unspecified. Counted at admission; sums to vllm:prefix_cache_hits + vllm:external_prefix_cache_hits. +# TYPE vllm:prompt_tokens_cached_by_source_created gauge +vllm:prompt_tokens_cached_by_source_created{engine="0",model_name="Llama-3.2-1B-Instruct",source="device"} 1.7929490428906596e+09 +vllm:prompt_tokens_cached_by_source_created{engine="0",model_name="Llama-3.2-1B-Instruct",source="host"} 1.7929490428906872e+09 +vllm:prompt_tokens_cached_by_source_created{engine="0",model_name="Llama-3.2-1B-Instruct",source="p2p"} 1.7929490428907108e+09 +vllm:prompt_tokens_cached_by_source_created{engine="0",model_name="Llama-3.2-1B-Instruct",source="disk"} 1.79294904289073e+09 +vllm:prompt_tokens_cached_by_source_created{engine="0",model_name="Llama-3.2-1B-Instruct",source="external_unspecified"} 1.7929490428907495e+09 +# HELP vllm:generation_tokens_total Number of generation tokens processed. +# TYPE vllm:generation_tokens_total counter +vllm:generation_tokens_total{engine="0",model_name="Llama-3.2-1B-Instruct"} 217.0 +# HELP vllm:generation_tokens_created Number of generation tokens processed. +# TYPE vllm:generation_tokens_created gauge +vllm:generation_tokens_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428907921e+09 +# HELP vllm:request_success_total Count of successfully processed requests. +# TYPE vllm:request_success_total counter +vllm:request_success_total{engine="0",finished_reason="stop",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_success_total{engine="0",finished_reason="length",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_success_total{engine="0",finished_reason="abort",model_name="Llama-3.2-1B-Instruct"} 0.0 +vllm:request_success_total{engine="0",finished_reason="error",model_name="Llama-3.2-1B-Instruct"} 0.0 +vllm:request_success_total{engine="0",finished_reason="repetition",model_name="Llama-3.2-1B-Instruct"} 0.0 +# HELP vllm:request_success_created Count of successfully processed requests. +# TYPE vllm:request_success_created gauge +vllm:request_success_created{engine="0",finished_reason="stop",model_name="Llama-3.2-1B-Instruct"} 1.792949042890855e+09 +vllm:request_success_created{engine="0",finished_reason="length",model_name="Llama-3.2-1B-Instruct"} 1.792949042890876e+09 +vllm:request_success_created{engine="0",finished_reason="abort",model_name="Llama-3.2-1B-Instruct"} 1.7929490428908954e+09 +vllm:request_success_created{engine="0",finished_reason="error",model_name="Llama-3.2-1B-Instruct"} 1.7929490428909163e+09 +vllm:request_success_created{engine="0",finished_reason="repetition",model_name="Llama-3.2-1B-Instruct"} 1.7929490428909378e+09 +# HELP vllm:admission_rejections_total Number of admission attempts rejected by a queue limit, by reason. +# TYPE vllm:admission_rejections_total counter +vllm:admission_rejections_total{model_name="Llama-3.2-1B-Instruct",reason="max_num_queued_reqs"} 0.0 +vllm:admission_rejections_total{model_name="Llama-3.2-1B-Instruct",reason="max_num_queued_tokens"} 0.0 +# HELP vllm:admission_rejections_created Number of admission attempts rejected by a queue limit, by reason. +# TYPE vllm:admission_rejections_created gauge +vllm:admission_rejections_created{model_name="Llama-3.2-1B-Instruct",reason="max_num_queued_reqs"} 1.7929490428910022e+09 +vllm:admission_rejections_created{model_name="Llama-3.2-1B-Instruct",reason="max_num_queued_tokens"} 1.7929490428910196e+09 +# HELP vllm:request_prompt_tokens Number of prefill tokens processed. +# TYPE vllm:request_prompt_tokens histogram +vllm:request_prompt_tokens_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 0.0 +vllm:request_prompt_tokens_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 0.0 +vllm:request_prompt_tokens_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 0.0 +vllm:request_prompt_tokens_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prompt_tokens_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prompt_tokens_bucket{engine="0",le="50.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_bucket{engine="0",le="100.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_bucket{engine="0",le="200.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_bucket{engine="0",le="500.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_bucket{engine="0",le="1000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_bucket{engine="0",le="2000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_bucket{engine="0",le="5000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_bucket{engine="0",le="10000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_bucket{engine="0",le="20000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_bucket{engine="0",le="50000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_bucket{engine="0",le="100000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prompt_tokens_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 82.0 +# HELP vllm:request_prompt_tokens_created Number of prefill tokens processed. +# TYPE vllm:request_prompt_tokens_created gauge +vllm:request_prompt_tokens_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428911083e+09 +# HELP vllm:request_generation_tokens Number of generation tokens processed. +# TYPE vllm:request_generation_tokens histogram +vllm:request_generation_tokens_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_generation_tokens_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_generation_tokens_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_generation_tokens_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_generation_tokens_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_generation_tokens_bucket{engine="0",le="50.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_bucket{engine="0",le="100.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_bucket{engine="0",le="200.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_bucket{engine="0",le="500.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_bucket{engine="0",le="1000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_bucket{engine="0",le="2000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_bucket{engine="0",le="5000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_bucket{engine="0",le="10000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_bucket{engine="0",le="20000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_bucket{engine="0",le="50000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_bucket{engine="0",le="100000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_generation_tokens_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 214.0 +# HELP vllm:request_generation_tokens_created Number of generation tokens processed. +# TYPE vllm:request_generation_tokens_created gauge +vllm:request_generation_tokens_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428912816e+09 +# HELP vllm:iteration_tokens_total Histogram of number of tokens per engine_step. +# TYPE vllm:iteration_tokens_total histogram +vllm:iteration_tokens_total_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 206.0 +vllm:iteration_tokens_total_bucket{engine="0",le="8.0",model_name="Llama-3.2-1B-Instruct"} 214.0 +vllm:iteration_tokens_total_bucket{engine="0",le="16.0",model_name="Llama-3.2-1B-Instruct"} 214.0 +vllm:iteration_tokens_total_bucket{engine="0",le="32.0",model_name="Llama-3.2-1B-Instruct"} 214.0 +vllm:iteration_tokens_total_bucket{engine="0",le="64.0",model_name="Llama-3.2-1B-Instruct"} 215.0 +vllm:iteration_tokens_total_bucket{engine="0",le="128.0",model_name="Llama-3.2-1B-Instruct"} 215.0 +vllm:iteration_tokens_total_bucket{engine="0",le="256.0",model_name="Llama-3.2-1B-Instruct"} 216.0 +vllm:iteration_tokens_total_bucket{engine="0",le="512.0",model_name="Llama-3.2-1B-Instruct"} 216.0 +vllm:iteration_tokens_total_bucket{engine="0",le="1024.0",model_name="Llama-3.2-1B-Instruct"} 216.0 +vllm:iteration_tokens_total_bucket{engine="0",le="2048.0",model_name="Llama-3.2-1B-Instruct"} 216.0 +vllm:iteration_tokens_total_bucket{engine="0",le="4096.0",model_name="Llama-3.2-1B-Instruct"} 216.0 +vllm:iteration_tokens_total_bucket{engine="0",le="8192.0",model_name="Llama-3.2-1B-Instruct"} 216.0 +vllm:iteration_tokens_total_bucket{engine="0",le="16384.0",model_name="Llama-3.2-1B-Instruct"} 216.0 +vllm:iteration_tokens_total_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 216.0 +vllm:iteration_tokens_total_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 216.0 +vllm:iteration_tokens_total_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 549.0 +# HELP vllm:iteration_tokens_total_created Histogram of number of tokens per engine_step. +# TYPE vllm:iteration_tokens_total_created gauge +vllm:iteration_tokens_total_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428914008e+09 +# HELP vllm:request_max_num_generation_tokens Histogram of maximum number of requested generation tokens. +# TYPE vllm:request_max_num_generation_tokens histogram +vllm:request_max_num_generation_tokens_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="50.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="100.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="200.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="500.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="1000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="2000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="5000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="10000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="20000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="50000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="100000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_max_num_generation_tokens_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 214.0 +# HELP vllm:request_max_num_generation_tokens_created Histogram of maximum number of requested generation tokens. +# TYPE vllm:request_max_num_generation_tokens_created gauge +vllm:request_max_num_generation_tokens_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428915174e+09 +# HELP vllm:request_params_n Histogram of the n request parameter. +# TYPE vllm:request_params_n histogram +vllm:request_params_n_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_n_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_n_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_n_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_n_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_n_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_n_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_n_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +# HELP vllm:request_params_n_created Histogram of the n request parameter. +# TYPE vllm:request_params_n_created gauge +vllm:request_params_n_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428916268e+09 +# HELP vllm:request_params_max_tokens Histogram of the max_tokens request parameter. +# TYPE vllm:request_params_max_tokens histogram +vllm:request_params_max_tokens_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_params_max_tokens_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_params_max_tokens_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_params_max_tokens_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_params_max_tokens_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 1.0 +vllm:request_params_max_tokens_bucket{engine="0",le="50.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_params_max_tokens_bucket{engine="0",le="100.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_bucket{engine="0",le="200.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_bucket{engine="0",le="500.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_bucket{engine="0",le="1000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_bucket{engine="0",le="2000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_bucket{engine="0",le="5000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_bucket{engine="0",le="10000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_bucket{engine="0",le="20000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_bucket{engine="0",le="50000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_bucket{engine="0",le="100000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_params_max_tokens_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 257.0 +# HELP vllm:request_params_max_tokens_created Histogram of the max_tokens request parameter. +# TYPE vllm:request_params_max_tokens_created gauge +vllm:request_params_max_tokens_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428917158e+09 +# HELP vllm:time_to_first_token_seconds Histogram of time to first token in seconds. +# TYPE vllm:time_to_first_token_seconds histogram +vllm:time_to_first_token_seconds_bucket{engine="0",le="0.001",model_name="Llama-3.2-1B-Instruct"} 0.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="0.005",model_name="Llama-3.2-1B-Instruct"} 0.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="0.01",model_name="Llama-3.2-1B-Instruct"} 0.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="0.02",model_name="Llama-3.2-1B-Instruct"} 6.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="0.04",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="0.06",model_name="Llama-3.2-1B-Instruct"} 9.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="0.08",model_name="Llama-3.2-1B-Instruct"} 9.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="0.1",model_name="Llama-3.2-1B-Instruct"} 9.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="0.25",model_name="Llama-3.2-1B-Instruct"} 9.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="0.5",model_name="Llama-3.2-1B-Instruct"} 9.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="0.75",model_name="Llama-3.2-1B-Instruct"} 9.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 9.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="2.5",model_name="Llama-3.2-1B-Instruct"} 9.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 9.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="7.5",model_name="Llama-3.2-1B-Instruct"} 10.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 10.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 10.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="40.0",model_name="Llama-3.2-1B-Instruct"} 10.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="80.0",model_name="Llama-3.2-1B-Instruct"} 10.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="160.0",model_name="Llama-3.2-1B-Instruct"} 10.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="640.0",model_name="Llama-3.2-1B-Instruct"} 10.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="2560.0",model_name="Llama-3.2-1B-Instruct"} 10.0 +vllm:time_to_first_token_seconds_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 10.0 +vllm:time_to_first_token_seconds_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 10.0 +vllm:time_to_first_token_seconds_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 5.258065223693848 +# HELP vllm:time_to_first_token_seconds_created Histogram of time to first token in seconds. +# TYPE vllm:time_to_first_token_seconds_created gauge +vllm:time_to_first_token_seconds_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428918285e+09 +# HELP vllm:inter_token_latency_seconds Histogram of inter-token latency in seconds. +# TYPE vllm:inter_token_latency_seconds histogram +vllm:inter_token_latency_seconds_bucket{engine="0",le="0.01",model_name="Llama-3.2-1B-Instruct"} 205.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="0.025",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="0.05",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="0.075",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="0.1",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="0.15",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="0.2",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="0.3",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="0.4",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="0.5",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="0.75",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="2.5",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="7.5",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="40.0",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="80.0",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 207.0 +vllm:inter_token_latency_seconds_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.3429912109859288 +# HELP vllm:inter_token_latency_seconds_created Histogram of inter-token latency in seconds. +# TYPE vllm:inter_token_latency_seconds_created gauge +vllm:inter_token_latency_seconds_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428919754e+09 +# HELP vllm:request_time_per_output_token_seconds Histogram of time_per_output_token_seconds per request. +# TYPE vllm:request_time_per_output_token_seconds histogram +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="0.01",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="0.025",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="0.05",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="0.075",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="0.1",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="0.15",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="0.2",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="0.3",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="0.4",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="0.5",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="0.75",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="2.5",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="7.5",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="40.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="80.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_time_per_output_token_seconds_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.045322619641721705 +# HELP vllm:request_time_per_output_token_seconds_created Histogram of time_per_output_token_seconds per request. +# TYPE vllm:request_time_per_output_token_seconds_created gauge +vllm:request_time_per_output_token_seconds_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428921254e+09 +# HELP vllm:e2e_request_latency_seconds Histogram of e2e request latency in seconds. +# TYPE vllm:e2e_request_latency_seconds histogram +vllm:e2e_request_latency_seconds_bucket{engine="0",le="0.3",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="0.5",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="0.8",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="1.5",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="2.5",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="15.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="30.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="40.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="50.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="60.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="120.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="240.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="480.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="960.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="1920.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="7680.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:e2e_request_latency_seconds_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 6.5164635181427 +# HELP vllm:e2e_request_latency_seconds_created Histogram of e2e request latency in seconds. +# TYPE vllm:e2e_request_latency_seconds_created gauge +vllm:e2e_request_latency_seconds_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.792949042892261e+09 +# HELP vllm:request_queue_time_seconds Histogram of time spent in WAITING phase for request. +# TYPE vllm:request_queue_time_seconds histogram +vllm:request_queue_time_seconds_bucket{engine="0",le="0.3",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="0.5",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="0.8",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="1.5",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="2.5",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="15.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="30.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="40.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="50.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="60.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="120.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="240.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="480.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="960.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="1920.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="7680.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_queue_time_seconds_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.00017091090558096766 +# HELP vllm:request_queue_time_seconds_created Histogram of time spent in WAITING phase for request. +# TYPE vllm:request_queue_time_seconds_created gauge +vllm:request_queue_time_seconds_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428923926e+09 +# HELP vllm:request_inference_time_seconds Histogram of time spent in RUNNING phase for request. +# TYPE vllm:request_inference_time_seconds histogram +vllm:request_inference_time_seconds_bucket{engine="0",le="0.3",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="0.5",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="0.8",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="1.5",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="2.5",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="15.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="30.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="40.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="50.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="60.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="120.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="240.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="480.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="960.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="1920.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="7680.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_inference_time_seconds_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 6.4989423449733295 +# HELP vllm:request_inference_time_seconds_created Histogram of time spent in RUNNING phase for request. +# TYPE vllm:request_inference_time_seconds_created gauge +vllm:request_inference_time_seconds_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428925264e+09 +# HELP vllm:request_prefill_time_seconds Histogram of time spent in PREFILL phase for request. +# TYPE vllm:request_prefill_time_seconds histogram +vllm:request_prefill_time_seconds_bucket{engine="0",le="0.3",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="0.5",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="0.8",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="1.5",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="2.5",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="15.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="30.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="40.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="50.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="60.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="120.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="240.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="480.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="960.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="1920.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="7680.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_time_seconds_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 5.167134573974181 +# HELP vllm:request_prefill_time_seconds_created Histogram of time spent in PREFILL phase for request. +# TYPE vllm:request_prefill_time_seconds_created gauge +vllm:request_prefill_time_seconds_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428929203e+09 +# HELP vllm:request_decode_time_seconds Histogram of time spent in DECODE phase for request. +# TYPE vllm:request_decode_time_seconds histogram +vllm:request_decode_time_seconds_bucket{engine="0",le="0.3",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="0.5",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="0.8",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="1.5",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="2.5",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="15.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="30.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="40.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="50.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="60.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="120.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="240.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="480.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="960.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="1920.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="7680.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_decode_time_seconds_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.3318077709991485 +# HELP vllm:request_decode_time_seconds_created Histogram of time spent in DECODE phase for request. +# TYPE vllm:request_decode_time_seconds_created gauge +vllm:request_decode_time_seconds_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428930626e+09 +# HELP vllm:request_num_preemptions Histogram of the number of times a request was preempted. +# TYPE vllm:request_num_preemptions histogram +vllm:request_num_preemptions_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_num_preemptions_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_num_preemptions_bucket{engine="0",le="3.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_num_preemptions_bucket{engine="0",le="4.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_num_preemptions_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_num_preemptions_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_num_preemptions_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_num_preemptions_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_num_preemptions_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_num_preemptions_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 0.0 +# HELP vllm:request_num_preemptions_created Histogram of the number of times a request was preempted. +# TYPE vllm:request_num_preemptions_created gauge +vllm:request_num_preemptions_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428931959e+09 +# HELP vllm:request_prefill_kv_computed_tokens Histogram of new KV tokens computed during prefill (excluding cached tokens). +# TYPE vllm:request_prefill_kv_computed_tokens histogram +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="1.0",model_name="Llama-3.2-1B-Instruct"} 0.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="2.0",model_name="Llama-3.2-1B-Instruct"} 0.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="5.0",model_name="Llama-3.2-1B-Instruct"} 0.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="10.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="20.0",model_name="Llama-3.2-1B-Instruct"} 7.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="50.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="100.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="200.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="500.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="1000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="2000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="5000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="10000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="20000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="50000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="100000.0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_bucket{engine="0",le="+Inf",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_count{engine="0",model_name="Llama-3.2-1B-Instruct"} 8.0 +vllm:request_prefill_kv_computed_tokens_sum{engine="0",model_name="Llama-3.2-1B-Instruct"} 82.0 +# HELP vllm:request_prefill_kv_computed_tokens_created Histogram of new KV tokens computed during prefill (excluding cached tokens). +# TYPE vllm:request_prefill_kv_computed_tokens_created gauge +vllm:request_prefill_kv_computed_tokens_created{engine="0",model_name="Llama-3.2-1B-Instruct"} 1.7929490428932803e+09 +# HELP vllm:cache_config_info Information of the LLMEngine CacheConfig +# TYPE vllm:cache_config_info gauge +vllm:cache_config_info{_block_size_resolved="True",block_size="16",cache_dtype="auto",cache_hit_alignment_tokens="None",effective_attention_block_size="16",enable_mamba_shared_prefix_checkpoint="False",enable_prefix_caching="True",engine="0",gpu_memory_utilization="0.92",hash_block_size="None",is_attention_free="False",kv_cache_dtype_skip_layers="[]",kv_cache_layout="None",kv_cache_max_concurrency="19.34716796875",kv_cache_memory_bytes="None",kv_cache_size_tokens="2535872",kv_offloading_backend="native",kv_offloading_size="None",kv_sharing_fast_prefill="False",mamba_block_size="None",mamba_cache_dtype="auto",mamba_cache_mode="none",mamba_page_size_padded="None",mamba_ssm_cache_dtype="auto",num_cpu_blocks="None",num_gpu_blocks="158492",num_gpu_blocks_override="None",prefix_cache_retention_interval="0",prefix_caching_hash_algo="sha256",prefix_match_unit="None",replayssm_buffer_len="16",skip_page_size_padded="None",sliding_window="None",swa_bounded_replay="True",use_kda_recoverssm="False",use_replayssm="False",user_specified_block_size="False"} 1.0 +# HELP http_requests_total Total number of requests by method, status and handler. +# TYPE http_requests_total counter +http_requests_total{handler="/v1/chat/completions",method="POST",status="2xx"} 1.0 +http_requests_total{handler="/v1/completions",method="POST",status="2xx"} 7.0 +# HELP http_requests_created Total number of requests by method, status and handler. +# TYPE http_requests_created gauge +http_requests_created{handler="/v1/chat/completions",method="POST",status="2xx"} 1.7929491236093621e+09 +http_requests_created{handler="/v1/completions",method="POST",status="2xx"} 1.7929491238378942e+09 +# HELP http_request_size_bytes Content length of incoming requests by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. +# TYPE http_request_size_bytes summary +http_request_size_bytes_count{handler="/v1/chat/completions"} 1.0 +http_request_size_bytes_sum{handler="/v1/chat/completions"} 124.0 +http_request_size_bytes_count{handler="/v1/completions"} 7.0 +http_request_size_bytes_sum{handler="/v1/completions"} 629.0 +# HELP http_request_size_bytes_created Content length of incoming requests by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. +# TYPE http_request_size_bytes_created gauge +http_request_size_bytes_created{handler="/v1/chat/completions"} 1.792949123609402e+09 +http_request_size_bytes_created{handler="/v1/completions"} 1.7929491238379269e+09 +# HELP http_response_size_bytes Content length of outgoing responses by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. +# TYPE http_response_size_bytes summary +http_response_size_bytes_count{handler="/v1/chat/completions"} 1.0 +http_response_size_bytes_sum{handler="/v1/chat/completions"} 748.0 +http_response_size_bytes_count{handler="/v1/completions"} 7.0 +http_response_size_bytes_sum{handler="/v1/completions"} 4773.0 +# HELP http_response_size_bytes_created Content length of outgoing responses by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. +# TYPE http_response_size_bytes_created gauge +http_response_size_bytes_created{handler="/v1/chat/completions"} 1.7929491236094475e+09 +http_response_size_bytes_created{handler="/v1/completions"} 1.7929491238379674e+09 +# HELP http_request_duration_highr_seconds Latency with many buckets but no API specific labels. Made for more accurate percentile calculations. +# TYPE http_request_duration_highr_seconds histogram +http_request_duration_highr_seconds_bucket{le="0.01"} 0.0 +http_request_duration_highr_seconds_bucket{le="0.025"} 1.0 +http_request_duration_highr_seconds_bucket{le="0.05"} 1.0 +http_request_duration_highr_seconds_bucket{le="0.075"} 1.0 +http_request_duration_highr_seconds_bucket{le="0.1"} 1.0 +http_request_duration_highr_seconds_bucket{le="0.25"} 7.0 +http_request_duration_highr_seconds_bucket{le="0.5"} 7.0 +http_request_duration_highr_seconds_bucket{le="0.75"} 7.0 +http_request_duration_highr_seconds_bucket{le="1.0"} 7.0 +http_request_duration_highr_seconds_bucket{le="1.5"} 7.0 +http_request_duration_highr_seconds_bucket{le="2.0"} 7.0 +http_request_duration_highr_seconds_bucket{le="2.5"} 7.0 +http_request_duration_highr_seconds_bucket{le="3.0"} 7.0 +http_request_duration_highr_seconds_bucket{le="3.5"} 7.0 +http_request_duration_highr_seconds_bucket{le="4.0"} 7.0 +http_request_duration_highr_seconds_bucket{le="4.5"} 7.0 +http_request_duration_highr_seconds_bucket{le="5.0"} 7.0 +http_request_duration_highr_seconds_bucket{le="7.5"} 8.0 +http_request_duration_highr_seconds_bucket{le="10.0"} 8.0 +http_request_duration_highr_seconds_bucket{le="30.0"} 8.0 +http_request_duration_highr_seconds_bucket{le="60.0"} 8.0 +http_request_duration_highr_seconds_bucket{le="+Inf"} 8.0 +http_request_duration_highr_seconds_count 8.0 +http_request_duration_highr_seconds_sum 6.5360731040127575 +# HELP http_request_duration_highr_seconds_created Latency with many buckets but no API specific labels. Made for more accurate percentile calculations. +# TYPE http_request_duration_highr_seconds_created gauge +http_request_duration_highr_seconds_created 1.792949076616627e+09 +# HELP http_request_duration_seconds Latency with only few buckets by handler. Made to be only used if aggregation by handler is important. +# TYPE http_request_duration_seconds histogram +http_request_duration_seconds_bucket{handler="/v1/chat/completions",le="0.1",method="POST"} 0.0 +http_request_duration_seconds_bucket{handler="/v1/chat/completions",le="0.5",method="POST"} 0.0 +http_request_duration_seconds_bucket{handler="/v1/chat/completions",le="1.0",method="POST"} 0.0 +http_request_duration_seconds_bucket{handler="/v1/chat/completions",le="+Inf",method="POST"} 1.0 +http_request_duration_seconds_count{handler="/v1/chat/completions",method="POST"} 1.0 +http_request_duration_seconds_sum{handler="/v1/chat/completions",method="POST"} 5.187623290985357 +http_request_duration_seconds_bucket{handler="/v1/completions",le="0.1",method="POST"} 1.0 +http_request_duration_seconds_bucket{handler="/v1/completions",le="0.5",method="POST"} 7.0 +http_request_duration_seconds_bucket{handler="/v1/completions",le="1.0",method="POST"} 7.0 +http_request_duration_seconds_bucket{handler="/v1/completions",le="+Inf",method="POST"} 7.0 +http_request_duration_seconds_count{handler="/v1/completions",method="POST"} 7.0 +http_request_duration_seconds_sum{handler="/v1/completions",method="POST"} 1.3484498130274005 +# HELP http_request_duration_seconds_created Latency with only few buckets by handler. Made to be only used if aggregation by handler is important. +# TYPE http_request_duration_seconds_created gauge +http_request_duration_seconds_created{handler="/v1/chat/completions",method="POST"} 1.7929491236095016e+09 +http_request_duration_seconds_created{handler="/v1/completions",method="POST"} 1.792949123838015e+09 diff --git a/tests/vec_inf/client/test_helper.py b/tests/vec_inf/client/test_helper.py index cdea799c..499a8d52 100644 --- a/tests/vec_inf/client/test_helper.py +++ b/tests/vec_inf/client/test_helper.py @@ -1180,6 +1180,150 @@ def test_fetch_metrics_request_exception( assert "Connection refused" in result +FIXTURES_DIR = Path(__file__).parent / "fixtures" + + +class TestPrometheusMetricsParsing: + """Parsing of real vLLM 0.32 and SGLang Prometheus exposition text.""" + + @pytest.fixture + def collector(self, mock_status_response): + """Return a collector whose status lookup is mocked out.""" + with patch("vec_inf.client._helper.ModelStatusMonitor") as monitor: + monitor.return_value.process_model_status.return_value = ( + mock_status_response + ) + yield PerformanceMetricsCollector("12345") + + def test_parse_vllm_0_32_live_capture(self, collector): + """Parse /metrics captured from a live vLLM 0.32.0 server. + + The fixture was scraped from Llama-3.2-1B-Instruct served by vec-inf on + two A40 nodes (pipeline parallel 2) while 48 requests were in flight. + """ + text = (FIXTURES_DIR / "vllm_0.32.0_metrics.txt").read_text() + + parsed = collector._parse_metrics(text) + + assert parsed["requests_running"] == 5.0 + assert parsed["requests_waiting"] == 0.0 + assert parsed["kv_cache_usage"] == pytest.approx(0.0001261901306698343) + assert parsed["preemptions_total"] == 0.0 + assert parsed["total_prompt_tokens"] == 572.0 + assert parsed["total_generation_tokens"] == 217.0 + assert parsed["request_latency_sum"] == pytest.approx(6.5164635181427) + assert parsed["request_latency_count"] == 8.0 + assert parsed["queue_time_count"] == 8.0 + # stop (1) + length (7); abort, error and repetition are all 0 here + assert parsed["successful_requests_total"] == 8.0 + assert parsed["prefix_cache_queries"] == 1307.0 + assert parsed["prefix_cache_hits"] == 960.0 + assert parsed["prefix_cache_hit_rate"] == pytest.approx(960.0 / 1307.0) + # The removed V0 metrics are not in the output, so they are not reported + assert "gpu_cache_usage" not in parsed + assert "requests_swapped" not in parsed + + def test_parse_vllm_data_parallel_engines(self, collector): + """Samples from several engines are summed, KV cache usage is averaged.""" + text = """\ +# HELP vllm:num_requests_running Number of requests in model execution batches. +# TYPE vllm:num_requests_running gauge +vllm:num_requests_running{engine="0",model_name="m"} 3.0 +vllm:num_requests_running{engine="1",model_name="m"} 2.0 +vllm:num_requests_waiting{engine="0",model_name="m"} 1.0 +vllm:num_requests_waiting{engine="1",model_name="m"} 4.0 +vllm:kv_cache_usage_perc{engine="0",model_name="m"} 0.2 +vllm:kv_cache_usage_perc{engine="1",model_name="m"} 0.6 +# TYPE vllm:prompt_tokens counter +vllm:prompt_tokens_total{engine="0",model_name="m"} 100.0 +vllm:prompt_tokens_total{engine="1",model_name="m"} 50.0 +vllm:prompt_tokens_created{engine="0",model_name="m"} 1.7929e+09 +vllm:num_preemptions_total{engine="0",model_name="m"} 1.0 +vllm:num_preemptions_total{engine="1",model_name="m"} 2.0 +vllm:prefix_cache_queries_total{engine="0",model_name="m"} 100.0 +vllm:prefix_cache_queries_total{engine="1",model_name="m"} 300.0 +vllm:prefix_cache_hits_total{engine="0",model_name="m"} 50.0 +vllm:prefix_cache_hits_total{engine="1",model_name="m"} 50.0 +vllm:request_success_total{engine="0",finished_reason="stop",model_name="m"} 5.0 +vllm:request_success_total{engine="0",finished_reason="length",model_name="m"} 2.0 +vllm:request_success_total{engine="0",finished_reason="abort",model_name="m"} 7.0 +vllm:request_success_total{engine="1",finished_reason="error",model_name="m"} 9.0 +vllm:request_success_total{engine="1",finished_reason="repetition",model_name="m"} 1.0 +vllm:e2e_request_latency_seconds_sum{engine="0",model_name="m"} 4.0 1792949107000 +vllm:e2e_request_latency_seconds_count{engine="0",model_name="m"} 2.0 +vllm:e2e_request_latency_seconds_sum{engine="1",model_name="m"} 2.0 +vllm:e2e_request_latency_seconds_count{engine="1",model_name="m"} 2.0 +""" + parsed = collector._parse_metrics(text) + + assert parsed["requests_running"] == 5.0 + assert parsed["requests_waiting"] == 5.0 + assert parsed["kv_cache_usage"] == pytest.approx(0.4) + assert parsed["total_prompt_tokens"] == 150.0 + assert parsed["preemptions_total"] == 3.0 + assert parsed["prefix_cache_hit_rate"] == pytest.approx(100.0 / 400.0) + # abort and error are excluded, repetition counts as finished + assert parsed["successful_requests_total"] == 8.0 + # trailing exposition timestamps are ignored + assert parsed["request_latency_sum"] == 6.0 + assert parsed["request_latency_count"] == 4.0 + + def test_parse_prefix_cache_without_queries(self, collector): + """A fresh server reports 0 queries; the hit rate is 0, not a crash.""" + text = """\ +vllm:prefix_cache_queries_total{engine="0",model_name="m"} 0.0 +vllm:prefix_cache_hits_total{engine="0",model_name="m"} 0.0 +""" + parsed = collector._parse_metrics(text) + + assert parsed["prefix_cache_hit_rate"] == 0.0 + + def test_parse_label_values_with_spaces_and_quotes(self, collector): + """Label values may contain spaces, commas and escaped quotes.""" + text = ( + 'vllm:request_success_total{engine="0",finished_reason="abort",' + 'model_name="my model, \\"v2\\""} 3.0\n' + 'vllm:request_success_total{engine="0",finished_reason="stop",' + 'model_name="my model, \\"v2\\""} 4.0\n' + 'vllm:num_requests_running{engine="0",model_name="a b"} 2.0\n' + ) + parsed = collector._parse_metrics(text) + + assert parsed["successful_requests_total"] == 4.0 + assert parsed["requests_running"] == 2.0 + + def test_parse_sglang_metrics(self, collector): + """Parse SGLang metrics (names from SGLang 0.5.12.post1).""" + text = """\ +# HELP sglang:num_running_reqs The number of running requests. +# TYPE sglang:num_running_reqs gauge +sglang:num_running_reqs{model_name="m",pp_rank="0",tp_rank="0"} 3.0 +sglang:num_queue_reqs{model_name="m",pp_rank="0",tp_rank="0"} 2.0 +sglang:token_usage{model_name="m",pp_rank="0",tp_rank="0"} 0.25 +sglang:cache_hit_rate{model_name="m",pp_rank="0",tp_rank="0"} 0.5 +sglang:num_retracted_requests_total{model_name="m",pp_rank="0",tp_rank="0"} 1.0 +sglang:prompt_tokens_total{model_name="m"} 1000.0 +sglang:generation_tokens_total{model_name="m"} 400.0 +sglang:num_requests_total{model_name="m"} 10.0 +sglang:e2e_request_latency_seconds_sum{model_name="m"} 5.0 +sglang:e2e_request_latency_seconds_count{model_name="m"} 10.0 +sglang:queue_time_seconds_sum{model_name="m"} 0.5 +sglang:queue_time_seconds_count{model_name="m"} 10.0 +""" + parsed = collector._parse_metrics(text) + + assert parsed["requests_running"] == 3.0 + assert parsed["requests_waiting"] == 2.0 + assert parsed["kv_cache_usage"] == 0.25 + assert parsed["prefix_cache_hit_rate"] == 0.5 + assert parsed["preemptions_total"] == 1.0 + assert parsed["total_prompt_tokens"] == 1000.0 + assert parsed["total_generation_tokens"] == 400.0 + assert parsed["successful_requests_total"] == 10.0 + assert parsed["request_latency_sum"] == 5.0 + assert parsed["queue_time_sum"] == 0.5 + + class TestModelRegistry: """Unit tests for the ModelRegistry class.""" diff --git a/tests/vec_inf/client/test_slurm_script_generator.py b/tests/vec_inf/client/test_slurm_script_generator.py index 4e8a364f..dfdb5f7d 100644 --- a/tests/vec_inf/client/test_slurm_script_generator.py +++ b/tests/vec_inf/client/test_slurm_script_generator.py @@ -1,5 +1,6 @@ """Unit tests for slurm script generator component in the vec_inf.client module.""" +import subprocess import tempfile from pathlib import Path from unittest.mock import mock_open, patch @@ -244,6 +245,49 @@ def test_generate_launch_cmd_sglang_multinode(self, basic_params): assert "--context-length 8192" in launch_cmd assert "--tensor-parallel-size 4" in launch_cmd + def test_generate_launch_cmd_vllm_executor_backend( + self, basic_params, multinode_params + ): + """Only multi-node vLLM selects the Ray executor started by the script.""" + single = SlurmScriptGenerator(basic_params)._generate_launch_cmd() + multi = SlurmScriptGenerator(multinode_params)._generate_launch_cmd() + + assert "--distributed-executor-backend" not in single + assert multi.rstrip().endswith("--distributed-executor-backend ray") + # The command must not end with a dangling line continuation + assert not single.rstrip().endswith("\\") + + @pytest.mark.parametrize("num_nodes", ["1", "2"]) + @pytest.mark.parametrize("engine_args", [{}, {"--context-length": "8192"}]) + def test_generate_launch_cmd_sglang_enables_metrics( + self, basic_params, num_nodes, engine_args + ): + """SGLang is always launched with --enable-metrics so /metrics exists.""" + sglang_params = basic_params.copy() + sglang_params["engine"] = "sglang" + sglang_params["num_nodes"] = num_nodes + sglang_params["engine_args"] = engine_args + + generator = SlurmScriptGenerator(sglang_params) + launch_cmd = generator._generate_launch_cmd() + + assert launch_cmd.count("--enable-metrics") == 1 + # The whole generated script must still be valid bash + result = subprocess.run( + ["bash", "-n"], + input=generator._generate_script_content(), + text=True, + capture_output=True, + check=False, + ) + assert result.returncode == 0, result.stderr + + def test_generate_launch_cmd_vllm_has_no_enable_metrics(self, basic_params): + """VLLM always serves /metrics; the SGLang flag must not leak into it.""" + launch_cmd = SlurmScriptGenerator(basic_params)._generate_launch_cmd() + + assert "--enable-metrics" not in launch_cmd + def test_generate_server_setup_sglang_multinode(self, basic_params): """Test SGLang multi-node setup.""" sglang_params = basic_params.copy() @@ -723,6 +767,8 @@ def test_generate_model_launch_script_sglang( # Verify the script content includes SGLang command call_args = mock_write_text.call_args[0][0] assert "sglang.launch_server" in call_args + assert "--enable-metrics" in call_args + assert not call_args.rstrip().endswith("\\") @patch("pathlib.Path.touch") @patch("pathlib.Path.write_text") diff --git a/uv.lock b/uv.lock index c3a67312..c9b7e40f 100644 --- a/uv.lock +++ b/uv.lock @@ -10416,15 +10416,6 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/e3/4d/0275b1865ea81d32286ab8ba20a8eca95c518d918cff1a723eb001c8c2ab/sgl_deep_gemm-0.1.0-py3-none-manylinux2014_x86_64.whl", hash = "sha256:6f75113c1dbc6a15b6e7714135b52464015dae4497440b4dd05676f346a15724", size = 4500439, upload-time = "2026-05-13T05:39:17.008Z" }, ] -[[package]] -name = "sgl-kernel" -version = "0.3.21" -source = { registry = "https://pypi.org/simple" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/eb/2b/f1aeca98bc856c14d870f1dcf38bca35cf84ffe58874c67402b0f862ed18/sgl_kernel-0.3.21-cp310-abi3-manylinux2014_aarch64.whl", hash = "sha256:bafdcc26e9ce1e9102b99e4186d652fefbafe5c22ea2cbb5ffe07b331e83be1f", size = 626572173, upload-time = "2026-01-15T03:48:32.615Z" }, - { url = "https://files.pythonhosted.org/packages/36/9f/f836e126002c7cfcfe35418f6cff5a63fe3f529c609b334ca4775354b4d5/sgl_kernel-0.3.21-cp310-abi3-manylinux2014_x86_64.whl", hash = "sha256:57dfb3a2a3cd759f499c32e2bad5f6489b7c58f7f9a84ee00c53ec92d303aaab", size = 535601072, upload-time = "2026-01-14T05:41:11.594Z" }, -] - [[package]] name = "sglang" version = "0.5.12.post1" @@ -12019,7 +12010,6 @@ dependencies = [ [package.optional-dependencies] sglang = [ { name = "orjson" }, - { name = "sgl-kernel" }, { name = "sglang" }, { name = "torchao" }, ] @@ -12078,7 +12068,6 @@ requires-dist = [ { name = "ray", extras = ["default"], marker = "extra == 'vllm'", specifier = ">=2.51.0" }, { name = "requests", specifier = ">=2.31.0" }, { name = "rich", specifier = ">=13.7.0" }, - { name = "sgl-kernel", marker = "extra == 'sglang'", specifier = ">=0.3.0" }, { name = "sglang", marker = "extra == 'sglang'", specifier = ">=0.5.5" }, { name = "torchao", marker = "extra == 'sglang'", specifier = ">=0.9.0" }, { name = "torchcodec", marker = "extra == 'vllm'", specifier = ">=0.14.0" }, From e9f5f90fded7d19673c995b9f7af9e7f0e4557e8 Mon Sep 17 00:00:00 2001 From: "pre-commit-ci[bot]" <66853113+pre-commit-ci[bot]@users.noreply.github.com> Date: Sun, 25 Oct 2026 17:32:49 +0000 Subject: [PATCH 04/10] [pre-commit.ci] Add auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --- .../client/fixtures/vllm_0.32.0_metrics.txt | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/tests/vec_inf/client/fixtures/vllm_0.32.0_metrics.txt b/tests/vec_inf/client/fixtures/vllm_0.32.0_metrics.txt index 020a966d..0045a3ea 100644 --- a/tests/vec_inf/client/fixtures/vllm_0.32.0_metrics.txt +++ b/tests/vec_inf/client/fixtures/vllm_0.32.0_metrics.txt @@ -585,27 +585,27 @@ http_requests_total{handler="/v1/completions",method="POST",status="2xx"} 7.0 # TYPE http_requests_created gauge http_requests_created{handler="/v1/chat/completions",method="POST",status="2xx"} 1.7929491236093621e+09 http_requests_created{handler="/v1/completions",method="POST",status="2xx"} 1.7929491238378942e+09 -# HELP http_request_size_bytes Content length of incoming requests by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. +# HELP http_request_size_bytes Content length of incoming requests by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. # TYPE http_request_size_bytes summary http_request_size_bytes_count{handler="/v1/chat/completions"} 1.0 http_request_size_bytes_sum{handler="/v1/chat/completions"} 124.0 http_request_size_bytes_count{handler="/v1/completions"} 7.0 http_request_size_bytes_sum{handler="/v1/completions"} 629.0 -# HELP http_request_size_bytes_created Content length of incoming requests by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. +# HELP http_request_size_bytes_created Content length of incoming requests by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. # TYPE http_request_size_bytes_created gauge http_request_size_bytes_created{handler="/v1/chat/completions"} 1.792949123609402e+09 http_request_size_bytes_created{handler="/v1/completions"} 1.7929491238379269e+09 -# HELP http_response_size_bytes Content length of outgoing responses by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. +# HELP http_response_size_bytes Content length of outgoing responses by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. # TYPE http_response_size_bytes summary http_response_size_bytes_count{handler="/v1/chat/completions"} 1.0 http_response_size_bytes_sum{handler="/v1/chat/completions"} 748.0 http_response_size_bytes_count{handler="/v1/completions"} 7.0 http_response_size_bytes_sum{handler="/v1/completions"} 4773.0 -# HELP http_response_size_bytes_created Content length of outgoing responses by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. +# HELP http_response_size_bytes_created Content length of outgoing responses by handler. Only value of header is respected. Otherwise ignored. No percentile calculated. # TYPE http_response_size_bytes_created gauge http_response_size_bytes_created{handler="/v1/chat/completions"} 1.7929491236094475e+09 http_response_size_bytes_created{handler="/v1/completions"} 1.7929491238379674e+09 -# HELP http_request_duration_highr_seconds Latency with many buckets but no API specific labels. Made for more accurate percentile calculations. +# HELP http_request_duration_highr_seconds Latency with many buckets but no API specific labels. Made for more accurate percentile calculations. # TYPE http_request_duration_highr_seconds histogram http_request_duration_highr_seconds_bucket{le="0.01"} 0.0 http_request_duration_highr_seconds_bucket{le="0.025"} 1.0 @@ -631,10 +631,10 @@ http_request_duration_highr_seconds_bucket{le="60.0"} 8.0 http_request_duration_highr_seconds_bucket{le="+Inf"} 8.0 http_request_duration_highr_seconds_count 8.0 http_request_duration_highr_seconds_sum 6.5360731040127575 -# HELP http_request_duration_highr_seconds_created Latency with many buckets but no API specific labels. Made for more accurate percentile calculations. +# HELP http_request_duration_highr_seconds_created Latency with many buckets but no API specific labels. Made for more accurate percentile calculations. # TYPE http_request_duration_highr_seconds_created gauge http_request_duration_highr_seconds_created 1.792949076616627e+09 -# HELP http_request_duration_seconds Latency with only few buckets by handler. Made to be only used if aggregation by handler is important. +# HELP http_request_duration_seconds Latency with only few buckets by handler. Made to be only used if aggregation by handler is important. # TYPE http_request_duration_seconds histogram http_request_duration_seconds_bucket{handler="/v1/chat/completions",le="0.1",method="POST"} 0.0 http_request_duration_seconds_bucket{handler="/v1/chat/completions",le="0.5",method="POST"} 0.0 @@ -648,7 +648,7 @@ http_request_duration_seconds_bucket{handler="/v1/completions",le="1.0",method=" http_request_duration_seconds_bucket{handler="/v1/completions",le="+Inf",method="POST"} 7.0 http_request_duration_seconds_count{handler="/v1/completions",method="POST"} 7.0 http_request_duration_seconds_sum{handler="/v1/completions",method="POST"} 1.3484498130274005 -# HELP http_request_duration_seconds_created Latency with only few buckets by handler. Made to be only used if aggregation by handler is important. +# HELP http_request_duration_seconds_created Latency with only few buckets by handler. Made to be only used if aggregation by handler is important. # TYPE http_request_duration_seconds_created gauge http_request_duration_seconds_created{handler="/v1/chat/completions",method="POST"} 1.7929491236095016e+09 http_request_duration_seconds_created{handler="/v1/completions",method="POST"} 1.792949123838015e+09 From c7b0546d0c807f8656d6babcda52ec6d4cfc259f Mon Sep 17 00:00:00 2001 From: Amrit Krishnan Date: Sun, 25 Oct 2026 14:17:01 -0400 Subject: [PATCH 05/10] Fix multi-node container launches, trust remote code for three VLMs, upgrade notes - Multi-node vllm runs Ray and vllm serve in separate apptainer instances. Ray 2.56 finds the local raylet by scanning the process table, so with --containall (private PID namespace) vllm serve failed with "No node info found matching attributes". Use --contain --cleanenv for the multi-node vllm containers; single node and SGLang keep --containall. Verified on Bon Echo with two A40 nodes. - Phi-3-vision-128k-instruct, Phi-3.5-vision-instruct and Molmo-7B-D-0924 ship custom code; vllm rejects their configs without --trust-remote-code (same on 0.22), so add it to models.yaml. - Add a live SGLang 0.5.12.post1 /metrics capture as a test fixture. - Add vllm 0.32 upgrade notes to the user guide. --- docs/user_guide.md | 13 + .../fixtures/sglang_0.5.12.post1_metrics.txt | 484 ++++++++++++++++++ tests/vec_inf/client/test_helper.py | 21 + .../client/test_slurm_script_generator.py | 19 + vec_inf/client/_slurm_script_generator.py | 8 +- vec_inf/client/_slurm_templates.py | 8 + vec_inf/config/models.yaml | 8 + 7 files changed, 559 insertions(+), 2 deletions(-) create mode 100644 tests/vec_inf/client/fixtures/sglang_0.5.12.post1_metrics.txt diff --git a/docs/user_guide.md b/docs/user_guide.md index 3d4b3cd3..cc7f85d9 100644 --- a/docs/user_guide.md +++ b/docs/user_guide.md @@ -387,6 +387,19 @@ ssh -L 8081:10.1.1.29:8081 username@v.vectorinstitute.ai -N ``` The example provided above is for the Vector Killarney cluster, change the variables accordingly for your environment. The IP address for the compute nodes on Killarney follow `10.1.1.XX` pattern, where `XX` is the GPU number (`kn029` -> `29` in this example). Similarly, for Bon Echo it's `172.17.8.XX`, where `XX` is from `gpuXX`. +## Notes on the vLLM 0.32 upgrade + +The vLLM image moved from vLLM 0.22.0 to 0.32.0. Things you may notice: + +* **Python, CUDA and drivers**: vLLM 0.32 needs Python 3.11 or newer, so the `vllm` extra skips vLLM on Python 3.10. The image uses torch 2.13 built for CUDA 13.0 on a CUDA 13.0.3 base image, which needs NVIDIA driver 580 or newer on the GPU nodes. +* **Your own environment (`--venv`)**: FlashInfer compiles some kernels the first time a server starts and needs a CUDA 13.0 toolkit whose `nvcc` and headers match (set `CUDA_HOME`). Without one, startup fails with "Could not find nvcc"; pass `--env VLLM_USE_FLASHINFER_SAMPLER=0` to avoid the compile. The container image already has the toolkit. +* **Removed model architectures** (since 0.22): Aquila, Arctic, AyaVision, Baichuan, Bamba, Chameleon, Cheers, Dots1, ERNIE (embedding and classification), Fairseq2Llama, FireRedLID, Fuyu, GritLM, Grok-1, HCXVision, the `InternLMForCausalLM` alias, JAIS (`JAISLMHeadModel`), Mantis, MiniMax-Text-01, MiniMax-M1 and MiniMax-VL-01, Mono-InternVL (`InternLM2VEForCausalLM`), MPT, MusicFlamingo, Ouro, Persimmon, Plamo2, PrithviGeoSpatialMAE, first generation Qwen and Qwen-VL (`QWenLMHeadModel`), Tarsier and Tarsier2, TeleChat, Xverse, and the `RWForCausalLM` and `StableLMEpochForCausalLM` aliases. None of the models in the default `models.yaml` use them. +* **Plugins**: GGUF and bitsandbytes checkpoints now need out-of-tree vLLM plugins that are not in the image. GPTQ act-order (`g_idx`) is ignored. +* **Defaults**: Model Runner V2 is used for all models, `--max-num-batched-tokens` defaults to 16384 (was 8192), and the server allows no CORS origins unless you pass `--allowed-origins`. `max_tokens` is now only an upper bound and is no longer reserved from the context window. +* **API responses**: chat responses use `reasoning` instead of `reasoning_content`, and an empty `tool_calls` list is left out. +* **Flags**: `-O3` means `--optimization-level 3`. The removed flags `--max-num-partial-prefills`, `--max-long-partial-prefills`, `--calculate-kv-scales` and `--override-attention-dtype` now fail to parse, and `--disable-cascade-attn` only prints a warning. +* **`metrics` command**: see the [`metrics` command](#metrics-command) section for the new rows. + ## Python API Usage You can also use the `vec_inf` Python API to launch and manage inference servers. diff --git a/tests/vec_inf/client/fixtures/sglang_0.5.12.post1_metrics.txt b/tests/vec_inf/client/fixtures/sglang_0.5.12.post1_metrics.txt new file mode 100644 index 00000000..fc258747 --- /dev/null +++ b/tests/vec_inf/client/fixtures/sglang_0.5.12.post1_metrics.txt @@ -0,0 +1,484 @@ +# HELP sglang:cuda_graph_passes_total Total number of forward passes categorized by CUDA graph. +# TYPE sglang:cuda_graph_passes_total counter +sglang:cuda_graph_passes_total{engine_type="unified",mode="prefill_cuda_graph",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:cuda_graph_passes_total{engine_type="unified",mode="decode_cuda_graph",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 184.0 +# HELP sglang:realtime_tokens_total Total number of tokens processed (updated on each log interval). mode: prefill_compute, prefill_cache, decode. +# TYPE sglang:realtime_tokens_total counter +sglang:realtime_tokens_total{engine_type="unified",mode="prefill_compute",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 48.0 +sglang:realtime_tokens_total{engine_type="unified",mode="decode",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 184.0 +sglang:realtime_tokens_total{engine_type="unified",mode="prefill_cache",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 18.0 +# HELP sglang:num_retracted_reqs The number of retracted requests. +# TYPE sglang:num_retracted_reqs gauge +sglang:num_retracted_reqs{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pid="44",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:num_paused_reqs The number of paused requests by async weight sync. +# TYPE sglang:num_paused_reqs gauge +sglang:num_paused_reqs{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pid="44",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:process_cpu_seconds_total Total CPU time consumed by this process (user + system) +# TYPE sglang:process_cpu_seconds_total counter +sglang:process_cpu_seconds_total{component="tokenizer"} 0.7999999999999998 +sglang:process_cpu_seconds_total{component="detokenizer"} 0.06000000000000094 +# HELP sglang:http_requests_total Total number of HTTP requests by endpoint and method +# TYPE sglang:http_requests_total counter +sglang:http_requests_total{endpoint="/model_info",method="GET"} 1.0 +sglang:http_requests_total{endpoint="/generate",method="POST"} 1.0 +sglang:http_requests_total{endpoint="/health",method="GET"} 7.0 +sglang:http_requests_total{endpoint="/v1/chat/completions",method="POST"} 1.0 +sglang:http_requests_total{endpoint="/v1/completions",method="POST"} 5.0 +sglang:http_requests_total{endpoint="/metrics",method="GET"} 1.0 +# HELP sglang:http_responses_total Total number of HTTP responses by endpoint and status code +# TYPE sglang:http_responses_total counter +sglang:http_responses_total{endpoint="/model_info",method="GET",status_code="200"} 1.0 +sglang:http_responses_total{endpoint="/health",method="GET",status_code="503"} 6.0 +sglang:http_responses_total{endpoint="/generate",method="POST",status_code="200"} 1.0 +sglang:http_responses_total{endpoint="/health",method="GET",status_code="200"} 1.0 +sglang:http_responses_total{endpoint="/v1/chat/completions",method="POST",status_code="200"} 1.0 +sglang:http_responses_total{endpoint="/v1/completions",method="POST",status_code="200"} 5.0 +# HELP sglang:prompt_tokens_total Number of prefill tokens processed. +# TYPE sglang:prompt_tokens_total counter +sglang:prompt_tokens_total{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 65.0 +# HELP sglang:generation_tokens_total Number of generation tokens processed. +# TYPE sglang:generation_tokens_total counter +sglang:generation_tokens_total{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 183.0 +# HELP sglang:num_requests_total Number of requests processed. +# TYPE sglang:num_requests_total counter +sglang:num_requests_total{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +# HELP sglang:cached_tokens_total Number of cached prompt tokens by source (device/host/storage). +# TYPE sglang:cached_tokens_total counter +sglang:cached_tokens_total{cache_source="device",engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 18.0 +# HELP sglang:max_total_num_tokens Maximum total number of tokens in the KV cache pool. +# TYPE sglang:max_total_num_tokens gauge +sglang:max_total_num_tokens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 3.160696e+06 +# HELP sglang:engine_startup_time The time taken for the engine to start up. +# TYPE sglang:engine_startup_time gauge +sglang:engine_startup_time{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:engine_load_weights_time The time taken for the engine to load weights. +# TYPE sglang:engine_load_weights_time gauge +sglang:engine_load_weights_time{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:page_size KV cache page size in tokens. +# TYPE sglang:page_size gauge +sglang:page_size{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 1.0 +# HELP sglang:num_pages Number of KV cache pages. +# TYPE sglang:num_pages gauge +sglang:num_pages{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 3.160696e+06 +# HELP sglang:context_len Maximum context length. +# TYPE sglang:context_len gauge +sglang:context_len{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 32768.0 +# HELP sglang:startup_available_gpu_memory_gb Available GPU memory in GB at startup. +# TYPE sglang:startup_available_gpu_memory_gb gauge +sglang:startup_available_gpu_memory_gb{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 5.29229736328125 +# HELP sglang:num_running_reqs The number of running requests. +# TYPE sglang:num_running_reqs gauge +sglang:num_running_reqs{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 1.0 +# HELP sglang:num_queue_reqs The number of requests in the waiting queue. +# TYPE sglang:num_queue_reqs gauge +sglang:num_queue_reqs{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:num_grammar_queue_reqs The number of requests in the grammar waiting queue. +# TYPE sglang:num_grammar_queue_reqs gauge +sglang:num_grammar_queue_reqs{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:gen_throughput The generation throughput (token/s). +# TYPE sglang:gen_throughput gauge +sglang:gen_throughput{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 235.656418651589 +# HELP sglang:cache_hit_rate The prefix cache hit rate. +# TYPE sglang:cache_hit_rate gauge +sglang:cache_hit_rate{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:decode_sum_seq_lens The sum of all sequence lengths in decode. +# TYPE sglang:decode_sum_seq_lens gauge +sglang:decode_sum_seq_lens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 13.0 +# HELP sglang:token_usage The token usage. +# TYPE sglang:token_usage gauge +sglang:token_usage{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:full_token_usage The token usage for full attention layers. +# TYPE sglang:full_token_usage gauge +sglang:full_token_usage{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 4.429404156552861e-06 +# HELP sglang:swa_token_usage The token usage for SWA layers. +# TYPE sglang:swa_token_usage gauge +sglang:swa_token_usage{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:mamba_usage The token usage for Mamba layers. +# TYPE sglang:mamba_usage gauge +sglang:mamba_usage{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:num_used_tokens The number of used tokens. +# TYPE sglang:num_used_tokens gauge +sglang:num_used_tokens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 14.0 +# HELP sglang:kv_available_tokens Number of free token slots in the KV cache pool. +# TYPE sglang:kv_available_tokens gauge +sglang:kv_available_tokens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 3.160493e+06 +# HELP sglang:kv_evictable_tokens Number of evictable (radix-cached) token slots in the KV cache pool. +# TYPE sglang:kv_evictable_tokens gauge +sglang:kv_evictable_tokens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 189.0 +# HELP sglang:kv_used_tokens Number of actively used token slots in the KV cache pool. +# TYPE sglang:kv_used_tokens gauge +sglang:kv_used_tokens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 14.0 +# HELP sglang:swa_available_tokens Number of free token slots in the SWA pool (hybrid-SWA only). +# TYPE sglang:swa_available_tokens gauge +sglang:swa_available_tokens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:swa_evictable_tokens Number of evictable (radix-cached) token slots in the SWA pool. +# TYPE sglang:swa_evictable_tokens gauge +sglang:swa_evictable_tokens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:swa_used_tokens Number of actively used token slots in the SWA pool. +# TYPE sglang:swa_used_tokens gauge +sglang:swa_used_tokens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:mamba_available_tokens Number of free state slots in the mamba SSM pool (hybrid-SSM only). +# TYPE sglang:mamba_available_tokens gauge +sglang:mamba_available_tokens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:mamba_evictable_tokens Number of evictable (radix-cached) state slots in the mamba SSM pool. +# TYPE sglang:mamba_evictable_tokens gauge +sglang:mamba_evictable_tokens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:mamba_used_tokens Number of actively used state slots in the mamba SSM pool. +# TYPE sglang:mamba_used_tokens gauge +sglang:mamba_used_tokens{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:spec_accept_length Mean acceptance length of speculative decoding (accepted drafts + bonus token per forward). +# TYPE sglang:spec_accept_length gauge +sglang:spec_accept_length{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:spec_accept_rate Speculative acceptance rate (`accepted drafts / proposed drafts` in batch). +# TYPE sglang:spec_accept_rate gauge +sglang:spec_accept_rate{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:num_prefill_bootstrap_queue_reqs The number of requests in the prefill bootstrap queue. +# TYPE sglang:num_prefill_bootstrap_queue_reqs gauge +sglang:num_prefill_bootstrap_queue_reqs{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:num_prefill_inflight_queue_reqs The number of requests in the prefill inflight queue. +# TYPE sglang:num_prefill_inflight_queue_reqs gauge +sglang:num_prefill_inflight_queue_reqs{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:num_decode_prealloc_queue_reqs The number of requests in the decode prealloc queue. +# TYPE sglang:num_decode_prealloc_queue_reqs gauge +sglang:num_decode_prealloc_queue_reqs{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:num_decode_transfer_queue_reqs The number of requests in the decode transfer queue. +# TYPE sglang:num_decode_transfer_queue_reqs gauge +sglang:num_decode_transfer_queue_reqs{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:pending_prealloc_token_usage The token usage for pending preallocated tokens (not preallocated yet). +# TYPE sglang:pending_prealloc_token_usage gauge +sglang:pending_prealloc_token_usage{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:utilization The utilization. +# TYPE sglang:utilization gauge +sglang:utilization{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:fwd_occupancy Forward pass GPU occupancy percentage. +# TYPE sglang:fwd_occupancy gauge +sglang:fwd_occupancy{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} NaN +# HELP sglang:new_token_ratio The new token ratio. +# TYPE sglang:new_token_ratio gauge +sglang:new_token_ratio{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.7 +# HELP sglang:is_cuda_graph Whether the batch is using CUDA graph. +# TYPE sglang:is_cuda_graph gauge +sglang:is_cuda_graph{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:num_unique_running_routing_keys Number of unique routing keys in running batch. +# TYPE sglang:num_unique_running_routing_keys gauge +sglang:num_unique_running_routing_keys{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:routing_key_running_req_count Distribution of routing keys by running request count (gt < count <= le). +# TYPE sglang:routing_key_running_req_count gauge +sglang:routing_key_running_req_count{engine_type="unified",gt="0",le="1",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_running_req_count{engine_type="unified",gt="1",le="2",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_running_req_count{engine_type="unified",gt="2",le="3",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_running_req_count{engine_type="unified",gt="3",le="5",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_running_req_count{engine_type="unified",gt="5",le="7",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_running_req_count{engine_type="unified",gt="7",le="10",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_running_req_count{engine_type="unified",gt="10",le="20",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_running_req_count{engine_type="unified",gt="20",le="50",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_running_req_count{engine_type="unified",gt="50",le="100",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_running_req_count{engine_type="unified",gt="100",le="200",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_running_req_count{engine_type="unified",gt="200",le="+Inf",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:routing_key_all_req_count Distribution of routing keys by running+waiting request count (gt < count <= le). +# TYPE sglang:routing_key_all_req_count gauge +sglang:routing_key_all_req_count{engine_type="unified",gt="0",le="1",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_all_req_count{engine_type="unified",gt="1",le="2",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_all_req_count{engine_type="unified",gt="2",le="3",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_all_req_count{engine_type="unified",gt="3",le="5",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_all_req_count{engine_type="unified",gt="5",le="7",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_all_req_count{engine_type="unified",gt="7",le="10",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_all_req_count{engine_type="unified",gt="10",le="20",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_all_req_count{engine_type="unified",gt="20",le="50",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_all_req_count{engine_type="unified",gt="50",le="100",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_all_req_count{engine_type="unified",gt="100",le="200",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:routing_key_all_req_count{engine_type="unified",gt="200",le="+Inf",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +# HELP sglang:time_to_first_token_seconds Histogram of time to first token in seconds. +# TYPE sglang:time_to_first_token_seconds histogram +sglang:time_to_first_token_seconds_sum{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 213.55322927888483 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="0.1",model_name="Qwen2.5-0.5B-Instruct"} 0.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="0.2",model_name="Qwen2.5-0.5B-Instruct"} 4.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="0.4",model_name="Qwen2.5-0.5B-Instruct"} 4.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="0.6",model_name="Qwen2.5-0.5B-Instruct"} 4.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="0.8",model_name="Qwen2.5-0.5B-Instruct"} 4.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="1.0",model_name="Qwen2.5-0.5B-Instruct"} 4.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="2.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="4.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="6.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="8.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="10.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="20.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="40.0",model_name="Qwen2.5-0.5B-Instruct"} 6.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="60.0",model_name="Qwen2.5-0.5B-Instruct"} 6.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="80.0",model_name="Qwen2.5-0.5B-Instruct"} 6.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="100.0",model_name="Qwen2.5-0.5B-Instruct"} 6.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="200.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="400.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:time_to_first_token_seconds_bucket{engine_type="unified",le="+Inf",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:time_to_first_token_seconds_count{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +# HELP sglang:e2e_request_latency_seconds Histogram of End-to-end request latency in seconds +# TYPE sglang:e2e_request_latency_seconds histogram +sglang:e2e_request_latency_seconds_sum{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 213.5532864499255 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="0.1",model_name="Qwen2.5-0.5B-Instruct"} 0.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="0.2",model_name="Qwen2.5-0.5B-Instruct"} 4.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="0.4",model_name="Qwen2.5-0.5B-Instruct"} 4.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="0.6",model_name="Qwen2.5-0.5B-Instruct"} 4.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="0.8",model_name="Qwen2.5-0.5B-Instruct"} 4.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="1.0",model_name="Qwen2.5-0.5B-Instruct"} 4.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="2.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="4.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="6.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="8.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="10.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="20.0",model_name="Qwen2.5-0.5B-Instruct"} 5.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="40.0",model_name="Qwen2.5-0.5B-Instruct"} 6.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="60.0",model_name="Qwen2.5-0.5B-Instruct"} 6.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="80.0",model_name="Qwen2.5-0.5B-Instruct"} 6.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="100.0",model_name="Qwen2.5-0.5B-Instruct"} 6.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="200.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="400.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="600.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="1200.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="1800.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="2400.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:e2e_request_latency_seconds_bucket{engine_type="unified",le="+Inf",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:e2e_request_latency_seconds_count{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +# HELP sglang:prompt_tokens_histogram Histogram of prompt token length. +# TYPE sglang:prompt_tokens_histogram histogram +sglang:prompt_tokens_histogram_sum{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 65.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="100.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="300.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="700.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="1000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="1500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="2000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="3000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="4000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="5000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="6000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="7000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="8000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="9000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="10000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="12500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="15000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="17500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="20000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="22500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="25000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="27500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="30000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="35000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="40000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="60000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="80000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="100000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="200000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="300000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="400000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="600000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="800000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="1e+06",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="1.1e+06",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_bucket{engine_type="unified",le="+Inf",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:prompt_tokens_histogram_count{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +# HELP sglang:uncached_prompt_tokens_histogram Histogram of uncached (compute) prompt token length. +# TYPE sglang:uncached_prompt_tokens_histogram histogram +sglang:uncached_prompt_tokens_histogram_sum{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 47.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="100.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="300.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="700.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="1000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="1500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="2000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="3000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="4000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="5000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="6000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="7000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="8000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="9000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="10000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="12500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="15000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="17500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="20000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="22500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="25000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="27500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="30000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="35000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="40000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="60000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="80000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="100000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="200000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="300000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="400000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="600000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="800000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="1e+06",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="1.1e+06",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_bucket{engine_type="unified",le="+Inf",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:uncached_prompt_tokens_histogram_count{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +# HELP sglang:generation_tokens_histogram Histogram of generation token length. +# TYPE sglang:generation_tokens_histogram histogram +sglang:generation_tokens_histogram_sum{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 183.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="100.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="300.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="700.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="1000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="1500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="2000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="3000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="4000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="5000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="6000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="7000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="8000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="9000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="10000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="12500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="15000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="17500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="20000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="22500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="25000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="27500.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="30000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="35000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="40000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="60000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="80000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="100000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="200000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="300000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="400000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="600000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="800000.0",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="1e+06",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="1.1e+06",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_bucket{engine_type="unified",le="+Inf",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +sglang:generation_tokens_histogram_count{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct"} 7.0 +# HELP sglang:per_stage_req_latency_seconds The latency of each stage of requests. +# TYPE sglang:per_stage_req_latency_seconds histogram +sglang:per_stage_req_latency_seconds_sum{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 0.0005351569852791727 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.001",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.0016200000000000001",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.0026244000000000007",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.004251528",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.006887475360000002",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.011157710083200004",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.018075490334784006",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.029282294342350097",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.047437316834607154",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.07684845327206359",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.12449449430074302",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.20168108076720373",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.3267233508428701",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.5292918283654495",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.8574527619520282",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="1.389073474362286",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="2.2502990284669036",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="3.6454844261163837",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="5.905684770308542",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="9.567209327899839",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="15.49887911119774",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="25.10818416014034",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="40.67525833942735",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="65.89391850987231",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="106.74814798599316",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="172.93199973730896",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="280.1498395744405",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="453.84274011059364",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="735.2252389791618",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="1191.064887146242",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="+Inf",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_count{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="request_process",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_sum{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 212.71969870303292 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.001",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 0.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.0016200000000000001",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 0.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.0026244000000000007",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 0.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.004251528",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 0.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.006887475360000002",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 0.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.011157710083200004",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 0.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.018075490334784006",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 0.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.029282294342350097",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 5.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.047437316834607154",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 5.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.07684845327206359",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 5.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.12449449430074302",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 5.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.20168108076720373",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 5.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.3267233508428701",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 5.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.5292918283654495",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 5.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="0.8574527619520282",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 5.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="1.389073474362286",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 6.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="2.2502990284669036",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 6.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="3.6454844261163837",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 6.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="5.905684770308542",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 6.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="9.567209327899839",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 6.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="15.49887911119774",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 6.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="25.10818416014034",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 6.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="40.67525833942735",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 7.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="65.89391850987231",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 7.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="106.74814798599316",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 7.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="172.93199973730896",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 7.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="280.1498395744405",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="453.84274011059364",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="735.2252389791618",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="1191.064887146242",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_bucket{engine_type="unified",le="+Inf",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 8.0 +sglang:per_stage_req_latency_seconds_count{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",stage="prefill_forward",tp_rank="0"} 8.0 +# HELP sglang:queue_time_seconds Histogram of queueing time in seconds. +# TYPE sglang:queue_time_seconds histogram +sglang:queue_time_seconds_sum{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.006551449012476951 +sglang:queue_time_seconds_bucket{engine_type="unified",le="0.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 0.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="0.001",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 7.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="0.005",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="0.01",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="0.05",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="0.1",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="0.2",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="0.5",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="1.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="2.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="3.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="4.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="5.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="10.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="15.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="20.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="30.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="40.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="50.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="60.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="70.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="80.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="90.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="100.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="200.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="300.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="400.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="500.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="600.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="700.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="800.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="900.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="1000.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="1200.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="1400.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="1600.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="1800.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="2000.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="2500.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="3000.0",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_bucket{engine_type="unified",le="+Inf",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +sglang:queue_time_seconds_count{engine_type="unified",model_name="Qwen2.5-0.5B-Instruct",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 8.0 +# HELP sglang:routing_keys_active Number of unique routing keys with active requests +# TYPE sglang:routing_keys_active gauge +sglang:routing_keys_active 0.0 +# HELP sglang:http_requests_active Number of currently active HTTP requests +# TYPE sglang:http_requests_active gauge +sglang:http_requests_active{endpoint="/model_info",method="GET"} 0.0 +sglang:http_requests_active{endpoint="/generate",method="POST"} 0.0 +sglang:http_requests_active{endpoint="/health",method="GET"} 0.0 +sglang:http_requests_active{endpoint="/v1/chat/completions",method="POST"} 0.0 +sglang:http_requests_active{endpoint="/v1/completions",method="POST"} 0.0 +sglang:http_requests_active{endpoint="/metrics",method="GET"} 1.0 diff --git a/tests/vec_inf/client/test_helper.py b/tests/vec_inf/client/test_helper.py index 499a8d52..ebacf8e2 100644 --- a/tests/vec_inf/client/test_helper.py +++ b/tests/vec_inf/client/test_helper.py @@ -1223,6 +1223,27 @@ def test_parse_vllm_0_32_live_capture(self, collector): assert "gpu_cache_usage" not in parsed assert "requests_swapped" not in parsed + def test_parse_sglang_0_5_12_live_capture(self, collector): + """Parse /metrics captured from a live SGLang 0.5.12.post1 server. + + The fixture was scraped from Qwen2.5-0.5B-Instruct on one A40, started + with --enable-metrics, after a health check, one chat and five + completion requests. + """ + text = (FIXTURES_DIR / "sglang_0.5.12.post1_metrics.txt").read_text() + + parsed = collector._parse_metrics(text) + + assert parsed["total_prompt_tokens"] == 65.0 + assert parsed["total_generation_tokens"] == 183.0 + assert parsed["successful_requests_total"] == 7.0 + assert parsed["request_latency_count"] == 7.0 + assert parsed["queue_time_count"] == 8.0 + assert parsed["requests_running"] == 1.0 + assert parsed["requests_waiting"] == 0.0 + assert parsed["kv_cache_usage"] == 0.0 + assert parsed["prefix_cache_hit_rate"] == 0.0 + def test_parse_vllm_data_parallel_engines(self, collector): """Samples from several engines are summed, KV cache usage is averaged.""" text = """\ diff --git a/tests/vec_inf/client/test_slurm_script_generator.py b/tests/vec_inf/client/test_slurm_script_generator.py index dfdb5f7d..1a42f49e 100644 --- a/tests/vec_inf/client/test_slurm_script_generator.py +++ b/tests/vec_inf/client/test_slurm_script_generator.py @@ -257,6 +257,25 @@ def test_generate_launch_cmd_vllm_executor_backend( # The command must not end with a dangling line continuation assert not single.rstrip().endswith("\\") + def test_multinode_vllm_container_shares_pid_namespace(self, singularity_params): + """Ray and vllm serve containers must share the host PID namespace. + + Ray finds the local raylet by scanning the process table, so with + --containall (a private PID namespace per container) vllm serve cannot + attach to the Ray cluster started by the script. + """ + multinode = singularity_params.copy() + multinode["num_nodes"] = "2" + content = SlurmScriptGenerator(multinode)._generate_script_content() + + assert "--containall" not in content + # ray start on the head, ray start on the workers, and vllm serve + assert content.count("--contain --cleanenv") == 3 + + single = SlurmScriptGenerator(singularity_params)._generate_script_content() + assert "--containall" in single + assert "--contain --cleanenv" not in single + @pytest.mark.parametrize("num_nodes", ["1", "2"]) @pytest.mark.parametrize("engine_args", [{}, {"--context-length": "8192"}]) def test_generate_launch_cmd_sglang_enables_metrics( diff --git a/vec_inf/client/_slurm_script_generator.py b/vec_inf/client/_slurm_script_generator.py index 7b9a3952..ff2b8b86 100644 --- a/vec_inf/client/_slurm_script_generator.py +++ b/vec_inf/client/_slurm_script_generator.py @@ -133,7 +133,7 @@ def _generate_server_setup(self) -> str: if self.use_container: server_setup_str = server_setup_str.replace( "CONTAINER_PLACEHOLDER", - SLURM_SCRIPT_TEMPLATE["container_command"].format( + SLURM_SCRIPT_TEMPLATE["container_command_ray"].format( env_str=self.env_str, image_path=IMAGE_PATH[self.engine], ), @@ -176,8 +176,12 @@ def _generate_launch_cmd(self) -> str: launch_cmd = ["\n"] if self.use_container: + # Multi-node vLLM must see the Ray processes started above + container_key = ( + "container_command_ray" if self.is_multinode else "container_command" + ) launch_cmd.append( - SLURM_SCRIPT_TEMPLATE["container_command"].format( + SLURM_SCRIPT_TEMPLATE[container_key].format( # type: ignore[literal-required] env_str=self.env_str, image_path=IMAGE_PATH[self.engine], ) diff --git a/vec_inf/client/_slurm_templates.py b/vec_inf/client/_slurm_templates.py index e9537c51..546c551e 100644 --- a/vec_inf/client/_slurm_templates.py +++ b/vec_inf/client/_slurm_templates.py @@ -82,6 +82,8 @@ class SlurmScriptTemplate(TypedDict): Bind path environment variable for the container container_command : str Template for container execution command + container_command_ray : str + Template for container execution command for multi-node vLLM (Ray) activate_venv : str Template for virtual environment activation server_setup : ServerSetupConfig @@ -99,6 +101,7 @@ class SlurmScriptTemplate(TypedDict): imports: str bind_path: str container_command: str + container_command_ray: str activate_venv: str server_setup: ServerSetupConfig find_server_port: list[str] @@ -119,6 +122,11 @@ class SlurmScriptTemplate(TypedDict): "imports": "source {src_dir}/find_port.sh", "bind_path": f"export {CONTAINER_MODULE_NAME_UPPER}_BINDPATH=${CONTAINER_MODULE_NAME_UPPER}_BINDPATH,/dev,/tmp,{{work_dir}}/.vec-inf-cache:$HOME/.cache,{{model_weights_path}}{{additional_binds}}", "container_command": f"{CONTAINER_MODULE_NAME} exec --nv {{env_str}} --containall {{image_path}} \\", + # Multi-node vLLM runs the Ray head/worker and `vllm serve` in separate + # container instances on the same node. Ray (2.56) finds the local raylet by + # scanning the process table, so these containers must share the host PID + # and IPC namespaces: --contain --cleanenv instead of --containall. + "container_command_ray": f"{CONTAINER_MODULE_NAME} exec --nv {{env_str}} --contain --cleanenv {{image_path}} \\", "activate_venv": "source {venv}/bin/activate", "server_setup": { "single_node": [ diff --git a/vec_inf/config/models.yaml b/vec_inf/config/models.yaml index ef704ca8..690d1616 100644 --- a/vec_inf/config/models.yaml +++ b/vec_inf/config/models.yaml @@ -418,9 +418,11 @@ models: num_nodes: 1 vocab_size: 32064 vllm_args: + --trust-remote-code: true --tensor-parallel-size: 2 --max-model-len: 65536 sglang_args: + --trust-remote-code: true --tensor-parallel-size: 2 --context-length: 65536 Llama-3.1-Nemotron-70B-Instruct-HF: @@ -777,9 +779,11 @@ models: num_nodes: 1 vocab_size: 32064 vllm_args: + --trust-remote-code: true --tensor-parallel-size: 2 --max-model-len: 65536 sglang_args: + --trust-remote-code: true --tensor-parallel-size: 2 --context-length: 65536 glm-4v-9b: @@ -796,6 +800,10 @@ models: gpus_per_node: 1 num_nodes: 1 vocab_size: 152064 + vllm_args: + --trust-remote-code: true + sglang_args: + --trust-remote-code: true deepseek-vl2: model_family: deepseek-vl2 model_type: VLM From 6221efc35a78810c99294e789172ed7554846a2c Mon Sep 17 00:00:00 2001 From: Amrit Krishnan Date: Sun, 25 Oct 2026 14:51:09 -0400 Subject: [PATCH 06/10] Address review: JSON engine args, SGLang metrics, NCCL env, docs - Split --vllm-args/--sglang-args only on commas outside {}, [] and quotes, accept '-tp 2' as well as '-tp=2', and add the -dc (--diffusion-config) alias. - Quote engine argument values in generated scripts with shlex.quote so JSON reaches the engine intact. Values with $ are double-quoted and a leading ~/ is left unquoted, so shell expansion still works there. - SGLang: count scheduler gauges only from tensor/pipeline/expert rank 0 (they repeat per rank), and report num_requests_total as finished_requests_total (it includes aborted requests). - Multi-node vllm containers now get $NCCL_ENV_ARG, and the NCCL_IB_GID_INDEX default is a real parameter expansion (the template dropped the $, so the value was the literal text {NCCL_IB_GID_INDEX:-3}). - vllm extra is now >=0.32,<0.33 (still Python 3.11+); lock unchanged apart from the specifier. - sglang image canary also imports datasets, pyarrow, transformers and protobuf. - Docs: Python API metric key changes, SGLang metric semantics, supported GPUs and drivers on Bon Echo, quoting of engine args, and the Python 3.10 note for the vllm extra in the README. --- README.md | 2 + docs/user_guide.md | 30 +++++- pyproject.toml | 2 +- sglang.Dockerfile | 7 +- tests/vec_inf/cli/test_helper.py | 11 +++ tests/vec_inf/client/test_helper.py | 88 ++++++++++++++++- .../client/test_slurm_script_generator.py | 94 +++++++++++++++++++ uv.lock | 2 +- vec_inf/cli/_helper.py | 16 +++- vec_inf/client/_client_vars.py | 13 ++- vec_inf/client/_helper.py | 67 +++++++++++-- vec_inf/client/_slurm_script_generator.py | 38 +++++++- vec_inf/client/_slurm_templates.py | 13 ++- 13 files changed, 353 insertions(+), 30 deletions(-) diff --git a/README.md b/README.md index 560d76ef..6dbab4b7 100644 --- a/README.md +++ b/README.md @@ -19,6 +19,8 @@ If you are using the Vector cluster environment, and you don't need any customiz ```bash pip install vec-inf ``` +The `vec-inf` client supports Python 3.10 and newer. The optional `vllm` extra (`pip install "vec-inf[vllm]"`) installs vLLM 0.32, which needs Python 3.11 or newer: on Python 3.10 the extra installs its other packages but not vLLM. + Otherwise, we recommend using the provided [`vllm.Dockerfile`](vllm.Dockerfile) and [`sglang.Dockerfile`](sglang.Dockerfile) to set up your own environment with the package. The built images are published to **Google Artifact Registry** at: ``` diff --git a/docs/user_guide.md b/docs/user_guide.md index cc7f85d9..99327a63 100644 --- a/docs/user_guide.md +++ b/docs/user_guide.md @@ -265,8 +265,9 @@ And you will see the performance metrics streamed to your console, note that the * The metrics come from the server's Prometheus `/metrics` endpoint. Counters are summed over all engines (for example with `--data-parallel-size`), and KV cache usage is averaged over them. * KV cache usage is the share of the GPU KV cache in use. vLLM 0.32 no longer reports separate GPU and CPU cache usage or swapped requests, so those rows were replaced by **KV Cache Usage** and **Preemptions**. -* The prefix cache hit rate is the share of prompt tokens served from the prefix cache since the server started. Prefix caching is on by default in vLLM and SGLang, so the row is shown whenever the server reports it. -* **Successful Requests** counts requests that finished normally. Requests aborted by the client or failed with an error are not counted. +* The prefix cache hit rate shows whenever the server reports it (prefix caching is on by default in vLLM and SGLang). For vLLM it is the share of prompt tokens served from the prefix cache since the server started. For SGLang it is SGLang's own `cache_hit_rate` gauge, which covers recent batches only. +* For vLLM, **Successful Requests** counts requests that finished normally; requests aborted by the client or failed with an error are not counted. SGLang does not report how requests ended, so for SGLang servers the row is **Finished Requests** and includes aborted requests. +* SGLang reports scheduler metrics once per tensor, pipeline and expert parallel rank. Only rank 0 is counted so running and waiting requests are not multiplied by the number of ranks; data parallel ranks are added up. * SGLang servers launched by `vec-inf` are started with `--enable-metrics`, which SGLang needs before it serves `/metrics`. ### `shutdown` command @@ -399,6 +400,31 @@ The vLLM image moved from vLLM 0.22.0 to 0.32.0. Things you may notice: * **API responses**: chat responses use `reasoning` instead of `reasoning_content`, and an empty `tool_calls` list is left out. * **Flags**: `-O3` means `--optimization-level 3`. The removed flags `--max-num-partial-prefills`, `--max-long-partial-prefills`, `--calculate-kv-scales` and `--override-attention-dtype` now fail to parse, and `--disable-cascade-attn` only prints a warning. * **`metrics` command**: see the [`metrics` command](#metrics-command) section for the new rows. +* **Python API metric keys**: `VecInfClient.get_metrics()` returns different keys, because vLLM no longer exports the old metrics: + + | Old key | New key | + | ------- | ------- | + | `gpu_cache_usage` | `kv_cache_usage` | + | `cpu_cache_usage` | removed, vLLM has no separate CPU cache gauge | + | `requests_swapped` | removed, vLLM no longer swaps; the closest signal is `preemptions_total` | + | `gpu_prefix_cache_hit_rate` | `prefix_cache_hit_rate` (also `prefix_cache_hits` and `prefix_cache_queries`) | + | `cpu_prefix_cache_hit_rate` | removed | + | `successful_requests_total` (SGLang) | `finished_requests_total`; vLLM keeps `successful_requests_total` | + + New keys: `preemptions_total`, `queue_time_count` and `avg_queue_time`. The `enabled_prefix_caching` attribute of `PerformanceMetricsCollector` and of the CLI's `MetricsResponseFormatter` is gone. +* **Engine arguments**: values are now quoted in the generated Slurm script, so JSON values such as `--vllm-args '-sc={"method":"ngram","num_speculative_tokens":3}'` reach the engine unchanged. Commas inside `{}`, `[]` or quotes no longer split arguments, and `-tp 2` works like `-tp=2`. Values that contain `$` or start with `~/` are still expanded by the shell. + +### Supported GPUs + +The vLLM image uses CUDA 13.0, which needs NVIDIA driver 580 or newer and a GPU with compute capability 7.5 or higher (Turing or newer). On the Vector **Bon Echo** cluster all GPU partitions qualify: every GPU type runs driver 580.178.04 (checked October 2026), and vLLM 0.32 was tested serving models on A40 and RTX 6000 nodes. + +| Partitions | GPU | Compute capability | +| ---------- | --- | ------------------ | +| `a100_b1` to `a100_b5` | A100 80GB | 8.0 | +| `a40_b1` to `a40_b5`, `interac` | A40 | 8.6 | +| `rtx6000_b1` to `rtx6000_b5`, `interac` | Quadro RTX 6000 | 7.5 | + +Turing GPUs (RTX 6000, T4) have no bfloat16 support, so vLLM loads bfloat16 models in float16 there. Older GPUs (Volta V100 and earlier) are not supported by CUDA 13. ## Python API Usage diff --git a/pyproject.toml b/pyproject.toml index b4032ae5..1b0f848e 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -54,7 +54,7 @@ inference = [ # (the vec-inf client itself still supports 3.10). vLLM 0.32 needs torchcodec # 0.14 or newer for video decoding. vllm = [ - "vllm[audio,bench]==0.32.0; python_version >= '3.11'", + "vllm[audio,bench]>=0.32,<0.33; python_version >= '3.11'", "ray[default]>=2.51.0", "torchcodec>=0.14.0", ] diff --git a/sglang.Dockerfile b/sglang.Dockerfile index eb087f16..90f9b21d 100644 --- a/sglang.Dockerfile +++ b/sglang.Dockerfile @@ -68,8 +68,11 @@ RUN uv export --frozen --no-emit-project --no-hashes \ rm -rf /root/.cache/uv /tmp/* # Build-time canary: fail the build if the locked deps cannot be imported. -RUN python3.12 -c "import sglang, torch; \ - print('sglang', sglang.__version__, '/ torch', torch.__version__)" +RUN python3.12 -c "import sglang, datasets, pyarrow, transformers, torch; \ + from google import protobuf; \ + print('sglang', sglang.__version__, '/ datasets', datasets.__version__, \ + '/ pyarrow', pyarrow.__version__, '/ protobuf', protobuf.__version__, \ + '/ torch', torch.__version__, '/ torch CUDA', torch.version.cuda)" # Install a single, system NCCL (from NVIDIA CUDA repo in base image) RUN apt-get update && apt-get install -y --allow-change-held-packages\ diff --git a/tests/vec_inf/cli/test_helper.py b/tests/vec_inf/cli/test_helper.py index af35b136..456c0d44 100644 --- a/tests/vec_inf/cli/test_helper.py +++ b/tests/vec_inf/cli/test_helper.py @@ -345,6 +345,17 @@ def test_format_metrics_with_latency(self): assert rows["Avg Request Latency"] == "2.5 s" assert rows["Avg Queue Time"] == "0.12 s" + def test_format_metrics_sglang_finished_requests(self): + """SGLang only reports finished requests, shown under that name.""" + metrics = {"finished_requests_total": 7, "kv_cache_usage": 0.1} + + formatter = MetricsResponseFormatter(metrics) + formatter.format_metrics() + + rows = _table_rows(formatter.table) + assert rows["Finished Requests"] == "7 reqs" + assert "Successful Requests" not in rows + def test_format_failed_metrics(self): """Test formatting failed metrics with error message.""" error_message = "ERROR: Server not ready" diff --git a/tests/vec_inf/client/test_helper.py b/tests/vec_inf/client/test_helper.py index ebacf8e2..1bdbdb3f 100644 --- a/tests/vec_inf/client/test_helper.py +++ b/tests/vec_inf/client/test_helper.py @@ -157,6 +157,63 @@ def test_process_engine_args_vllm_short_aliases( "--max-model-len": "4096", } + @patch("vec_inf.client._helper.utils.load_config") + @patch("pathlib.Path.exists") + def test_process_engine_args_json_values( + self, mock_path_exists, mock_load_config, mock_configs + ): + """Commas inside JSON objects, lists and quotes do not split arguments.""" + mock_load_config.return_value = mock_configs + mock_path_exists.return_value = True + + launcher = ModelLauncher("test-model", {}) + vllm_args = launcher._process_engine_args( + '-sc={"method":"ngram","num_speculative_tokens":3,"prompt_lookup_max":4},' + '-cc={"cudagraph_capture_sizes":[1,2,4]},' + "--max-model-len=8192," + '-ac={"backend": "FLASH_ATTN"},' + "--chat-template-content-format='a,b'," + '-dc={"steps": 4}', + "vllm", + ) + + assert vllm_args == { + "--speculative-config": ( + '{"method":"ngram","num_speculative_tokens":3,"prompt_lookup_max":4}' + ), + "--compilation-config": '{"cudagraph_capture_sizes":[1,2,4]}', + "--max-model-len": "8192", + "--attention-config": '{"backend": "FLASH_ATTN"}', + "--chat-template-content-format": "'a,b'", + "--diffusion-config": '{"steps": 4}', + } + + @patch("vec_inf.client._helper.utils.load_config") + @patch("pathlib.Path.exists") + def test_process_engine_args_space_separated_values( + self, mock_path_exists, mock_load_config, mock_configs + ): + """'-tp 2' is handled like '-tp=2', and JSON may follow a space.""" + mock_load_config.return_value = mock_configs + mock_path_exists.return_value = True + + launcher = ModelLauncher("test-model", {}) + vllm_args = launcher._process_engine_args( + '-tp 2, --max-model-len 4096, -sc {"method": "ngram"}, -ep', "vllm" + ) + + assert vllm_args == { + "--tensor-parallel-size": "2", + "--max-model-len": "4096", + "--speculative-config": '{"method": "ngram"}', + "--enable-expert-parallel": True, + } + sglang_args = launcher._process_engine_args("--tp 4, --ep-size 2", "sglang") + assert sglang_args == { + "--tensor-parallel-size": "4", + "--expert-parallel-size": "2", + } + @patch("vec_inf.client._helper.utils.load_config") @patch("pathlib.Path.exists") def test_process_engine_args_value_with_equals( @@ -1236,7 +1293,8 @@ def test_parse_sglang_0_5_12_live_capture(self, collector): assert parsed["total_prompt_tokens"] == 65.0 assert parsed["total_generation_tokens"] == 183.0 - assert parsed["successful_requests_total"] == 7.0 + assert parsed["finished_requests_total"] == 7.0 + assert "successful_requests_total" not in parsed assert parsed["request_latency_count"] == 7.0 assert parsed["queue_time_count"] == 8.0 assert parsed["requests_running"] == 1.0 @@ -1244,6 +1302,32 @@ def test_parse_sglang_0_5_12_live_capture(self, collector): assert parsed["kv_cache_usage"] == 0.0 assert parsed["prefix_cache_hit_rate"] == 0.0 + def test_parse_sglang_per_rank_gauges(self, collector): + """SGLang scheduler gauges repeated per TP/PP rank are counted once. + + With pipeline parallelism (or --enable-metrics-for-all-schedulers) each + rank reports the same running and waiting requests. Data parallel ranks + serve different requests and are summed. + """ + text = """\ +sglang:num_running_reqs{dp_rank="0",model_name="m",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 4.0 +sglang:num_running_reqs{dp_rank="0",model_name="m",moe_ep_rank="0",pp_rank="1",tp_rank="0"} 4.0 +sglang:num_running_reqs{dp_rank="0",model_name="m",moe_ep_rank="0",pp_rank="0",tp_rank="1"} 4.0 +sglang:num_running_reqs{dp_rank="1",model_name="m",moe_ep_rank="0",pp_rank="0",tp_rank="0"} 3.0 +sglang:num_running_reqs{dp_rank="1",model_name="m",moe_ep_rank="0",pp_rank="1",tp_rank="0"} 3.0 +sglang:num_queue_reqs{model_name="m",pp_rank="0",tp_rank="0"} 2.0 +sglang:num_queue_reqs{model_name="m",pp_rank="1",tp_rank="0"} 2.0 +sglang:token_usage{model_name="m",pp_rank="0",tp_rank="0"} 0.5 +sglang:token_usage{model_name="m",pp_rank="1",tp_rank="0"} 0.1 +sglang:num_requests_total{model_name="m"} 9.0 +""" + parsed = collector._parse_metrics(text) + + assert parsed["requests_running"] == 7.0 + assert parsed["requests_waiting"] == 2.0 + assert parsed["kv_cache_usage"] == 0.5 + assert parsed["finished_requests_total"] == 9.0 + def test_parse_vllm_data_parallel_engines(self, collector): """Samples from several engines are summed, KV cache usage is averaged.""" text = """\ @@ -1340,7 +1424,7 @@ def test_parse_sglang_metrics(self, collector): assert parsed["preemptions_total"] == 1.0 assert parsed["total_prompt_tokens"] == 1000.0 assert parsed["total_generation_tokens"] == 400.0 - assert parsed["successful_requests_total"] == 10.0 + assert parsed["finished_requests_total"] == 10.0 assert parsed["request_latency_sum"] == 5.0 assert parsed["queue_time_sum"] == 0.5 diff --git a/tests/vec_inf/client/test_slurm_script_generator.py b/tests/vec_inf/client/test_slurm_script_generator.py index 1a42f49e..1d975f3e 100644 --- a/tests/vec_inf/client/test_slurm_script_generator.py +++ b/tests/vec_inf/client/test_slurm_script_generator.py @@ -257,6 +257,100 @@ def test_generate_launch_cmd_vllm_executor_backend( # The command must not end with a dangling line continuation assert not single.rstrip().endswith("\\") + @pytest.mark.parametrize("num_nodes", ["1", "2"]) + @pytest.mark.parametrize("engine", ["vllm", "sglang"]) + def test_engine_arg_values_survive_bash( + self, basic_params, num_nodes, engine, tmp_path + ): + """JSON and spaced values reach the engine exactly as configured. + + The generated script must pass ``bash -n``, and the engine command, + run through bash with ``echo`` in place of the engine, must print each + value unchanged (no stripped quotes, no word splitting). + """ + spec = '{"method":"ngram","num_speculative_tokens":3}' + params = basic_params.copy() + params["engine"] = engine + params["num_nodes"] = num_nodes + params["engine_args"] = { + "--speculative-config": spec, + "--chat-template": "/path/with space/template.jinja", + "--max-model-len": "8192", + "--enforce-eager": True, + } + generator = SlurmScriptGenerator(params) + content = generator._generate_script_content() + + syntax = subprocess.run( + ["bash", "-n"], input=content, text=True, capture_output=True, check=False + ) + assert syntax.returncode == 0, syntax.stderr + + # Re-run only the argument list through bash and print one per line + lines = generator._generate_launch_cmd().splitlines() + start = next(i for i, line in enumerate(lines) if "--served-model-name" in line) + arg_lines = [ + line + for line in lines[start:] + if line.strip().startswith("--") and "$" not in line + ] + script = "printf '%s\\n' " + " ".join( + line.strip().rstrip("\\").strip() for line in arg_lines + ) + echoed = subprocess.run( + ["bash", "-c", script], text=True, capture_output=True, check=True + ).stdout.splitlines() + + assert echoed[echoed.index("--speculative-config") + 1] == spec + assert ( + echoed[echoed.index("--chat-template") + 1] + == "/path/with space/template.jinja" + ) + assert echoed[echoed.index("--max-model-len") + 1] == "8192" + assert "--enforce-eager" in echoed + + def test_engine_arg_values_keep_shell_expansion(self, basic_params): + """Values with $VAR or a leading ~/ are still expanded by bash.""" + params = basic_params.copy() + params["engine_args"] = { + "--download-dir": "$SLURM_TMPDIR/weights", + "--chat-template": "~/templates/chat.jinja", + } + launch_cmd = SlurmScriptGenerator(params)._generate_launch_cmd() + + assert '--download-dir "$SLURM_TMPDIR/weights"' in launch_cmd + assert "--chat-template ~/templates/chat.jinja" in launch_cmd + echoed = subprocess.run( + [ + "bash", + "-c", + 'printf "%s\\n" "$SLURM_TMPDIR/weights" ~/templates/chat.jinja', + ], + env={"SLURM_TMPDIR": "/tmp/job1", "HOME": "/home/u"}, + text=True, + capture_output=True, + check=True, + ).stdout.splitlines() + assert echoed == ["/tmp/job1/weights", "/home/u/templates/chat.jinja"] + + def test_multinode_vllm_container_passes_nccl_env(self, singularity_params): + """NCCL_IB_* settings from the RDMA check reach the Ray containers.""" + multinode = singularity_params.copy() + multinode["num_nodes"] = "2" + content = SlurmScriptGenerator(multinode)._generate_script_content() + + container_lines = [ + line for line in content.splitlines() if "--contain --cleanenv" in line + ] + assert len(container_lines) == 3 + assert all("$NCCL_ENV_ARG" in line for line in container_lines) + # Unquoted so "--env NCCL_IB_GID_INDEX=3" splits into two arguments + assert '"$NCCL_ENV_ARG"' not in content + # The GID index default must be a parameter expansion, not literal text + assert "${NCCL_IB_GID_INDEX:-3}" in content + assert "${NCCL_IB_GID_INDEX:-0}" in content + assert "={NCCL_IB_GID_INDEX" not in content + def test_multinode_vllm_container_shares_pid_namespace(self, singularity_params): """Ray and vllm serve containers must share the host PID namespace. diff --git a/uv.lock b/uv.lock index c9b7e40f..8ddf3ca0 100644 --- a/uv.lock +++ b/uv.lock @@ -12071,7 +12071,7 @@ requires-dist = [ { name = "sglang", marker = "extra == 'sglang'", specifier = ">=0.5.5" }, { name = "torchao", marker = "extra == 'sglang'", specifier = ">=0.9.0" }, { name = "torchcodec", marker = "extra == 'vllm'", specifier = ">=0.14.0" }, - { name = "vllm", extras = ["audio", "bench"], marker = "python_full_version >= '3.11' and extra == 'vllm'", specifier = "==0.32.0" }, + { name = "vllm", extras = ["audio", "bench"], marker = "python_full_version >= '3.11' and extra == 'vllm'", specifier = ">=0.32,<0.33" }, ] provides-extras = ["vllm", "sglang"] diff --git a/vec_inf/cli/_helper.py b/vec_inf/cli/_helper.py index 65539bf2..91037361 100644 --- a/vec_inf/cli/_helper.py +++ b/vec_inf/cli/_helper.py @@ -422,10 +422,18 @@ def format_metrics(self) -> None: "Total Generation Tokens", f"{self.metrics.get('total_generation_tokens', 0):.0f} tokens", ) - self.table.add_row( - "Successful Requests", - f"{self.metrics.get('successful_requests_total', 0):.0f} reqs", - ) + # SGLang only reports finished requests (including aborted ones), + # vLLM reports them by finish reason so successes can be counted + if "finished_requests_total" in self.metrics: + self.table.add_row( + "Finished Requests", + f"{self.metrics['finished_requests_total']:.0f} reqs", + ) + else: + self.table.add_row( + "Successful Requests", + f"{self.metrics.get('successful_requests_total', 0):.0f} reqs", + ) class ListCmdDisplay: diff --git a/vec_inf/client/_client_vars.py b/vec_inf/client/_client_vars.py index 0f4e99b6..8b256566 100644 --- a/vec_inf/client/_client_vars.py +++ b/vec_inf/client/_client_vars.py @@ -13,6 +13,8 @@ Mapping of vLLM and SGLang Prometheus metrics to output keys and aggregation KEY_METRICS_EXCLUDED_LABELS : dict Label values to skip when aggregating a metric +SGLANG_PRIMARY_RANK_LABELS : tuple + SGLang rank labels whose non-zero ranks repeat rank 0's scheduler metrics SLURM_JOB_CONFIG_ARGS : dict Mapping of SLURM configuration arguments to their parameter names VLLM_SHORT_TO_LONG_MAP : dict @@ -62,7 +64,8 @@ "sglang:e2e_request_latency_seconds_count": ("request_latency_count", "sum"), "sglang:queue_time_seconds_sum": ("queue_time_sum", "sum"), "sglang:queue_time_seconds_count": ("queue_time_count", "sum"), - "sglang:num_requests_total": ("successful_requests_total", "sum"), + # Counts every finished request, including aborted ones + "sglang:num_requests_total": ("finished_requests_total", "sum"), "sglang:num_running_reqs": ("requests_running", "sum"), "sglang:num_queue_reqs": ("requests_waiting", "sum"), "sglang:num_retracted_requests_total": ("preemptions_total", "sum"), @@ -77,6 +80,13 @@ "vllm:request_success_total": {"finished_reason": {"abort", "error"}}, } +# SGLang reports scheduler metrics once per tensor, pipeline and expert parallel +# rank (with pipeline parallelism, or --enable-metrics-for-all-schedulers), and +# every rank sees the same requests. Only samples from rank 0 of these groups +# are counted, so running/waiting requests are not multiplied by the number of +# ranks. Data parallel ranks (dp_rank) serve different requests and are summed. +SGLANG_PRIMARY_RANK_LABELS = ("tp_rank", "pp_rank", "moe_ep_rank") + # Slurm job configuration arguments SLURM_JOB_CONFIG_ARGS = { "job-name": "model_name", @@ -114,6 +124,7 @@ "-dpe": "--data-parallel-external-lb", "-dpm": "--data-parallel-multi-port-external-lb", "-asc": "--api-server-count", + "-dc": "--diffusion-config", "-ep": "--enable-expert-parallel", "-sc": "--speculative-config", "-cc": "--compilation-config", diff --git a/vec_inf/client/_helper.py b/vec_inf/client/_helper.py index d85596e9..e2e08f9a 100644 --- a/vec_inf/client/_helper.py +++ b/vec_inf/client/_helper.py @@ -21,6 +21,7 @@ ENGINE_SHORT_TO_LONG_MAP, KEY_METRICS, KEY_METRICS_EXCLUDED_LABELS, + SGLANG_PRIMARY_RANK_LABELS, SRC_DIR, SUPPORTED_ENGINES, ) @@ -54,6 +55,45 @@ _PROM_LABEL_RE = re.compile(r'([a-zA-Z_][a-zA-Z0-9_]*)="((?:[^"\\]|\\.)*)"') +def _split_engine_args(arg_string: str) -> list[str]: + """Split a comma-separated engine argument string. + + Commas inside braces, brackets or quotes do not split, so JSON values such + as ``-sc={"method":"ngram","num_speculative_tokens":3}`` stay in one piece. + + Parameters + ---------- + arg_string : str + Comma-separated engine arguments + + Returns + ------- + list[str] + The individual arguments + """ + parts: list[str] = [] + current: list[str] = [] + depth = 0 + quote = "" + for char in arg_string: + if quote: + if char == quote: + quote = "" + elif char in "\"'": + quote = char + elif char in "{[": + depth += 1 + elif char in "}]": + depth = max(depth - 1, 0) + elif char == "," and depth == 0: + parts.append("".join(current)) + current = [] + continue + current.append(char) + parts.append("".join(current)) + return parts + + class ModelLauncher: """Helper class for handling inference server launch. @@ -177,7 +217,7 @@ def _process_engine_args( engine_args: dict[str, str | bool] = {} engine_arg_map = ENGINE_SHORT_TO_LONG_MAP[engine_choice] - for raw_arg in arg_string.split(","): + for raw_arg in _split_engine_args(arg_string): arg = raw_arg.strip() if not arg: continue @@ -189,10 +229,14 @@ def _process_engine_args( if not level: raise ValueError("-O requires an optimization level, e.g. -O3") engine_args[engine_arg_map["-O"]] = level - elif "=" in arg: - key, value = arg.split("=", 1) - key = key.strip() - engine_args[engine_arg_map.get(key, key)] = value.strip() + continue + # A flag and its value are separated by whichever of "=" or + # whitespace comes first, so both "-tp=2" and "-tp 2" work and a + # value may itself contain "=" or spaces (e.g. JSON). + match = re.match(r"^(\S+?)(?:=|\s+)(.*)$", arg, re.DOTALL) + if match: + key, value = match.group(1), match.group(2).strip() + engine_args[engine_arg_map.get(key, key)] = value else: engine_args[engine_arg_map.get(arg, arg)] = True return engine_args @@ -1056,11 +1100,16 @@ def _parse_metrics(self, metrics_text: str) -> dict[str, float]: if metric_name not in KEY_METRICS: continue + labels = self._parse_labels(label_text) if label_text else {} excluded = KEY_METRICS_EXCLUDED_LABELS.get(metric_name) - if excluded and label_text: - labels = self._parse_labels(label_text) - if any(labels.get(k) in values for k, values in excluded.items()): - continue + if excluded and any( + labels.get(k) in values for k, values in excluded.items() + ): + continue + if metric_name.startswith("sglang:") and any( + labels.get(k, "0") != "0" for k in SGLANG_PRIMARY_RANK_LABELS + ): + continue try: value = float(value_text) diff --git a/vec_inf/client/_slurm_script_generator.py b/vec_inf/client/_slurm_script_generator.py index ff2b8b86..568547dc 100644 --- a/vec_inf/client/_slurm_script_generator.py +++ b/vec_inf/client/_slurm_script_generator.py @@ -4,6 +4,7 @@ servers in both single-node and multi-node configurations. """ +import shlex from datetime import datetime from pathlib import Path from typing import Any @@ -17,6 +18,35 @@ from vec_inf.client._slurm_vars import CONTAINER_MODULE_NAME, IMAGE_PATH +def format_engine_arg_value(value: Any) -> str: + r"""Quote an engine argument value for the generated bash script. + + Values are single-quoted with ``shlex.quote`` so JSON (e.g. for + ``--speculative-config``) and values with spaces reach the engine + unchanged. Two cases keep shell expansion on purpose: a value containing + ``$`` is double-quoted (escaping ``\\``, ``"`` and backticks) so variables + such as ``$SLURM_TMPDIR`` still expand, and a leading ``~/`` is left + unquoted so it still expands to the home directory. + + Parameters + ---------- + value : Any + The argument value + + Returns + ------- + str + The value, quoted for bash + """ + text = str(value) + if "$" in text: + escaped = text.replace("\\", "\\\\").replace('"', '\\"').replace("`", "\\`") + return f'"{escaped}"' + if text.startswith("~/"): + return "~/" + shlex.quote(text[2:]) + return shlex.quote(text) + + class SlurmScriptGenerator: """A class to generate Slurm scripts for running inference servers. @@ -199,7 +229,7 @@ def _generate_launch_cmd(self) -> str: if isinstance(value, bool): launch_cmd.append(f" {arg} \\") else: - launch_cmd.append(f" {arg} {value} \\") + launch_cmd.append(f" {arg} {format_engine_arg_value(value)} \\") # Multi-node vLLM runs on the Ray cluster started in the server setup # section, so the Ray executor must be selected explicitly. vLLM only @@ -241,7 +271,9 @@ def _generate_multinode_sglang_launch_cmd(self) -> str: if isinstance(value, bool): engine_arg_str += f" {arg} \\\n" else: - engine_arg_str += f" {arg} {value} \\\n" + engine_arg_str += ( + f" {arg} {format_engine_arg_value(value)} \\\n" + ) return launch_cmd.replace( "SGLANG_ARGS_PLACEHOLDER", engine_arg_str.rstrip("\\\n") @@ -365,7 +397,7 @@ def _generate_model_launch_script(self, model_name: str) -> Path: if isinstance(value, bool): script_content.append(f" {arg} \\") else: - script_content.append(f" {arg} {value} \\") + script_content.append(f" {arg} {format_engine_arg_value(value)} \\") script_content[-1] = script_content[-1].rstrip(" \\") # Write the bash script to the log directory launch_script_path = self._write_to_log_dir( diff --git a/vec_inf/client/_slurm_templates.py b/vec_inf/client/_slurm_templates.py index 546c551e..3bde75f4 100644 --- a/vec_inf/client/_slurm_templates.py +++ b/vec_inf/client/_slurm_templates.py @@ -126,7 +126,10 @@ class SlurmScriptTemplate(TypedDict): # container instances on the same node. Ray (2.56) finds the local raylet by # scanning the process table, so these containers must share the host PID # and IPC namespaces: --contain --cleanenv instead of --containall. - "container_command_ray": f"{CONTAINER_MODULE_NAME} exec --nv {{env_str}} --contain --cleanenv {{image_path}} \\", + # $NCCL_ENV_ARG (set by the RDMA check in multinode_vllm, e.g. + # "--env NCCL_IB_GID_INDEX=3") is left unquoted on purpose so it splits into + # apptainer arguments; --cleanenv would otherwise drop the NCCL_IB_* exports. + "container_command_ray": f"{CONTAINER_MODULE_NAME} exec --nv {{env_str}} $NCCL_ENV_ARG --contain --cleanenv {{image_path}} \\", "activate_venv": "source {venv}/bin/activate", "server_setup": { "single_node": [ @@ -148,12 +151,12 @@ class SlurmScriptTemplate(TypedDict): " # Pick GID index based on link layer (IB vs RoCE)", ' if ibv_devinfo 2>/dev/null | grep -q "link_layer:.*Ethernet"; then', " # RoCEv2 typically needs a nonzero GID index; 3 is common, try 2 if your fabric uses it", - " export NCCL_IB_GID_INDEX={{NCCL_IB_GID_INDEX:-3}}", - ' export NCCL_ENV_ARG="--env NCCL_IB_GID_INDEX={{NCCL_IB_GID_INDEX:-3}}"', + " export NCCL_IB_GID_INDEX=${{NCCL_IB_GID_INDEX:-3}}", + ' export NCCL_ENV_ARG="--env NCCL_IB_GID_INDEX=${{NCCL_IB_GID_INDEX:-3}}"', " else", " # Native InfiniBand => GID 0", - " export NCCL_IB_GID_INDEX={{NCCL_IB_GID_INDEX:-0}}", - ' export NCCL_ENV_ARG="--env NCCL_IB_GID_INDEX={{NCCL_IB_GID_INDEX:-0}}"', + " export NCCL_IB_GID_INDEX=${{NCCL_IB_GID_INDEX:-0}}", + ' export NCCL_ENV_ARG="--env NCCL_IB_GID_INDEX=${{NCCL_IB_GID_INDEX:-0}}"', " fi", "fi", "\n# Start Ray head node", From 9fff675ece1035ea1f9428e90dd3480d2ef6417d Mon Sep 17 00:00:00 2001 From: Amrit Krishnan Date: Sun, 25 Oct 2026 14:57:19 -0400 Subject: [PATCH 07/10] Keep the Ray head port out of Ray's worker port range find_available_port could pick a GCS port in Ray's default worker port range (10002-19999); ray start on the workers then failed with "worker_ports is trying to use a port number ... used by other components". Seen once on Bon Echo during a 2-node launch. --- tests/vec_inf/client/test_slurm_script_generator.py | 6 ++++++ vec_inf/client/_slurm_templates.py | 5 ++++- 2 files changed, 10 insertions(+), 1 deletion(-) diff --git a/tests/vec_inf/client/test_slurm_script_generator.py b/tests/vec_inf/client/test_slurm_script_generator.py index 1d975f3e..0a8f2265 100644 --- a/tests/vec_inf/client/test_slurm_script_generator.py +++ b/tests/vec_inf/client/test_slurm_script_generator.py @@ -333,6 +333,12 @@ def test_engine_arg_values_keep_shell_expansion(self, basic_params): ).stdout.splitlines() assert echoed == ["/tmp/job1/weights", "/home/u/templates/chat.jinja"] + def test_multinode_vllm_ray_port_outside_worker_range(self, multinode_params): + """The Ray head port is picked above Ray's worker port range.""" + setup = SlurmScriptGenerator(multinode_params)._generate_server_setup() + + assert "head_node_port=$(find_available_port $head_node 20000 65535)" in setup + def test_multinode_vllm_container_passes_nccl_env(self, singularity_params): """NCCL_IB_* settings from the RDMA check reach the Ray containers.""" multinode = singularity_params.copy() diff --git a/vec_inf/client/_slurm_templates.py b/vec_inf/client/_slurm_templates.py index 3bde75f4..31be7c3e 100644 --- a/vec_inf/client/_slurm_templates.py +++ b/vec_inf/client/_slurm_templates.py @@ -160,7 +160,10 @@ class SlurmScriptTemplate(TypedDict): " fi", "fi", "\n# Start Ray head node", - "head_node_port=$(find_available_port $head_node 8080 65535)", + # Keep the Ray GCS port out of Ray's default worker port range + # (10002-19999); a GCS port in that range makes `ray start` fail + # with "worker_ports is trying to use a port number ...". + "head_node_port=$(find_available_port $head_node 20000 65535)", "ray_head=$head_node_ip:$head_node_port", 'echo "Ray Head IP: $ray_head"', 'echo "Starting HEAD at $head_node"', From 8c56fb600be3dd14f4e35f1825138759d40402ae Mon Sep 17 00:00:00 2001 From: Amrit Krishnan Date: Sun, 25 Oct 2026 15:23:53 -0400 Subject: [PATCH 08/10] Cap the server port at 55535 for SGLang's gRPC port SGLang 0.5.12 also binds a gRPC port at the server port + 10000 and exits with "SGLANG_GRPC_PORT (...) must be between 1 and 65535" when vec-inf picks a server port above 55535. Seen on Bon Echo. --- tests/vec_inf/client/test_slurm_script_generator.py | 11 +++++++++++ vec_inf/client/_slurm_templates.py | 7 +++++-- 2 files changed, 16 insertions(+), 2 deletions(-) diff --git a/tests/vec_inf/client/test_slurm_script_generator.py b/tests/vec_inf/client/test_slurm_script_generator.py index 0a8f2265..a03ff445 100644 --- a/tests/vec_inf/client/test_slurm_script_generator.py +++ b/tests/vec_inf/client/test_slurm_script_generator.py @@ -333,6 +333,17 @@ def test_engine_arg_values_keep_shell_expansion(self, basic_params): ).stdout.splitlines() assert echoed == ["/tmp/job1/weights", "/home/u/templates/chat.jinja"] + @pytest.mark.parametrize("engine", ["vllm", "sglang"]) + def test_server_port_leaves_room_for_sglang_grpc(self, basic_params, engine): + """SGLang binds server port + 10000 for gRPC, so cap the port at 55535.""" + params = basic_params.copy() + params["engine"] = engine + setup = SlurmScriptGenerator(params)._generate_server_setup() + + assert ( + "server_port_number=$(find_available_port $head_node 8080 55535)" in setup + ) + def test_multinode_vllm_ray_port_outside_worker_range(self, multinode_params): """The Ray head port is picked above Ray's worker port range.""" setup = SlurmScriptGenerator(multinode_params)._generate_server_setup() diff --git a/vec_inf/client/_slurm_templates.py b/vec_inf/client/_slurm_templates.py index 31be7c3e..d34a8afd 100644 --- a/vec_inf/client/_slurm_templates.py +++ b/vec_inf/client/_slurm_templates.py @@ -195,7 +195,9 @@ class SlurmScriptTemplate(TypedDict): ], }, "find_server_port": [ - "\nserver_port_number=$(find_available_port $head_node 8080 65535)", + # Max 55535: SGLang also binds a gRPC port at server port + 10000 and + # refuses to start if that is above 65535 + "\nserver_port_number=$(find_available_port $head_node 8080 55535)", 'server_address="http://${head_node}:${server_port_number}/v1"', ], "write_to_json": [ @@ -318,7 +320,8 @@ class BatchModelLaunchScriptTemplate(TypedDict): "server_address_setup": [ "source {src_dir}/find_port.sh", "head_node_ip=${{SLURMD_NODENAME}}", - "server_port_number=$(find_available_port $head_node_ip 8080 65535)", + # Max 55535: SGLang also binds a gRPC port at server port + 10000 + "server_port_number=$(find_available_port $head_node_ip 8080 55535)", 'server_address="http://${{head_node_ip}}:${{server_port_number}}/v1"\n', "echo $server_address\n", ], From 61cd8330eb0008d5cbdc1273515b3df0c2611150 Mon Sep 17 00:00:00 2001 From: Amrit Krishnan Date: Sun, 25 Oct 2026 15:50:55 -0400 Subject: [PATCH 09/10] Address re-review: multi-value flags, quoting, SGLang DP ranks, NCCL, ports - Values that do not start with { or [ are split into words and each word is quoted on its own, so flags that take several values (--cudagraph-capture-sizes 1 2 4 8, --served-model-name a b, ...) work again in both the space and = forms. JSON stays one word. - Only words with a real variable reference ($NAME, ${...}) or an escaped \$ are double-quoted; anything else is single-quoted, so price$5 and JSON containing $ stay literal. - The comma splitter only tracks quotes inside {} or [] and raises a ValueError for unbalanced braces, brackets or quotes instead of silently swallowing the following arguments. - SGLang: per metric and dp_rank keep the lowest (pp, tp, moe_ep) rank that reports, then sum over dp_rank. With --enable-dp-attention a DP group's reporting scheduler can have tp_rank != 0. - Without ibv_devices on the host, leave the transport to NCCL instead of forcing NCCL_IB_DISABLE=1 into the containers. - Server port range starts at 20000, above Ray's worker ports. - Docs: engine argument quoting rules, NCCL and port changes. --- docs/user_guide.md | 10 ++- tests/vec_inf/client/test_helper.py | 87 ++++++++++++++++++- .../client/test_slurm_script_generator.py | 53 +++++++++-- vec_inf/client/_client_vars.py | 7 +- vec_inf/client/_helper.py | 82 +++++++++++++++-- vec_inf/client/_slurm_script_generator.py | 67 +++++++++++--- vec_inf/client/_slurm_templates.py | 19 ++-- 7 files changed, 283 insertions(+), 42 deletions(-) diff --git a/docs/user_guide.md b/docs/user_guide.md index 99327a63..f1f3eb0b 100644 --- a/docs/user_guide.md +++ b/docs/user_guide.md @@ -267,7 +267,7 @@ And you will see the performance metrics streamed to your console, note that the * KV cache usage is the share of the GPU KV cache in use. vLLM 0.32 no longer reports separate GPU and CPU cache usage or swapped requests, so those rows were replaced by **KV Cache Usage** and **Preemptions**. * The prefix cache hit rate shows whenever the server reports it (prefix caching is on by default in vLLM and SGLang). For vLLM it is the share of prompt tokens served from the prefix cache since the server started. For SGLang it is SGLang's own `cache_hit_rate` gauge, which covers recent batches only. * For vLLM, **Successful Requests** counts requests that finished normally; requests aborted by the client or failed with an error are not counted. SGLang does not report how requests ended, so for SGLang servers the row is **Finished Requests** and includes aborted requests. -* SGLang reports scheduler metrics once per tensor, pipeline and expert parallel rank. Only rank 0 is counted so running and waiting requests are not multiplied by the number of ranks; data parallel ranks are added up. +* SGLang can report scheduler metrics from several tensor, pipeline and expert parallel ranks of the same data parallel group. Only the lowest reporting rank of each group is counted, so running and waiting requests are not multiplied by the number of ranks; data parallel groups are added up. * SGLang servers launched by `vec-inf` are started with `--enable-metrics`, which SGLang needs before it serves `/metrics`. ### `shutdown` command @@ -412,7 +412,13 @@ The vLLM image moved from vLLM 0.22.0 to 0.32.0. Things you may notice: | `successful_requests_total` (SGLang) | `finished_requests_total`; vLLM keeps `successful_requests_total` | New keys: `preemptions_total`, `queue_time_count` and `avg_queue_time`. The `enabled_prefix_caching` attribute of `PerformanceMetricsCollector` and of the CLI's `MetricsResponseFormatter` is gone. -* **Engine arguments**: values are now quoted in the generated Slurm script, so JSON values such as `--vllm-args '-sc={"method":"ngram","num_speculative_tokens":3}'` reach the engine unchanged. Commas inside `{}`, `[]` or quotes no longer split arguments, and `-tp 2` works like `-tp=2`. Values that contain `$` or start with `~/` are still expanded by the shell. +* **Engine arguments** (`--vllm-args`, `--sglang-args` and `*_args` in `models.yaml`): + * Commas inside `{}` or `[]` (and inside quoted strings within them) no longer split arguments, so JSON values such as `--vllm-args '-sc={"method":"ngram","num_speculative_tokens":3}'` reach the engine unchanged. Unbalanced braces, brackets or quotes are reported as an error. + * A flag and its value can be separated by `=` or a space: `-tp=2` and `-tp 2` are the same. + * Values are quoted in the generated Slurm script. A value starting with `{` or `[` is passed as one word. Other values are split into words, so flags that take several values work in both forms, for example `--cudagraph-capture-sizes=1 2 4 8` or `--served-model-name a b`. To pass one word that contains spaces, quote it: `--chat-template='/path/with space/t.jinja'`. + * A word that references a variable (`$NAME` or `${NAME}`) is still expanded by the shell, `\$` gives a literal dollar sign, and a leading `~/` still expands to your home directory. Any other `$` (for example `price$5`) is passed literally. +* **Multi-node networking**: multi-node vLLM containers now receive the NCCL settings from the launch script's RDMA check. On nodes with RDMA tools, RoCE links get `NCCL_IB_GID_INDEX=3` and InfiniBand links `0` (override with `--env NCCL_IB_GID_INDEX=...`). Without `ibv_devices` on the host the script no longer forces `NCCL_IB_DISABLE=1`; NCCL picks the transport itself and falls back to TCP sockets when there is no usable device. +* **Ports**: the server port is now picked between 20000 and 55535, and the Ray head port between 20000 and 65535. ### Supported GPUs diff --git a/tests/vec_inf/client/test_helper.py b/tests/vec_inf/client/test_helper.py index 1bdbdb3f..6981552c 100644 --- a/tests/vec_inf/client/test_helper.py +++ b/tests/vec_inf/client/test_helper.py @@ -21,6 +21,7 @@ ModelStatusMonitor, PerformanceMetricsCollector, ) +from vec_inf.client._slurm_script_generator import format_engine_arg_value from vec_inf.client.config import ModelConfig from vec_inf.client.models import ( ModelStatus, @@ -172,7 +173,6 @@ def test_process_engine_args_json_values( '-cc={"cudagraph_capture_sizes":[1,2,4]},' "--max-model-len=8192," '-ac={"backend": "FLASH_ATTN"},' - "--chat-template-content-format='a,b'," '-dc={"steps": 4}', "vllm", ) @@ -184,10 +184,70 @@ def test_process_engine_args_json_values( "--compilation-config": '{"cudagraph_capture_sizes":[1,2,4]}', "--max-model-len": "8192", "--attention-config": '{"backend": "FLASH_ATTN"}', - "--chat-template-content-format": "'a,b'", "--diffusion-config": '{"steps": 4}', } + @patch("vec_inf.client._helper.utils.load_config") + @patch("pathlib.Path.exists") + def test_process_engine_args_multi_value_flags( + self, mock_path_exists, mock_load_config, mock_configs + ): + """Flags taking several values keep all of them, in both forms.""" + mock_load_config.return_value = mock_configs + mock_path_exists.return_value = True + launcher = ModelLauncher("test-model", {}) + + for arg_string in ( + "--cudagraph-capture-sizes 1 2 4 8,--served-model-name a b", + "--cudagraph-capture-sizes=1 2 4 8,--served-model-name=a b", + ): + args = launcher._process_engine_args(arg_string, "vllm") + assert args == { + "--cudagraph-capture-sizes": "1 2 4 8", + "--served-model-name": "a b", + } + # Each value becomes separate words in the generated script + assert format_engine_arg_value(args["--cudagraph-capture-sizes"]) == ( + "1 2 4 8" + ) + assert format_engine_arg_value(args["--served-model-name"]) == "a b" + + @patch("vec_inf.client._helper.utils.load_config") + @patch("pathlib.Path.exists") + def test_process_engine_args_quotes_outside_json( + self, mock_path_exists, mock_load_config, mock_configs + ): + """A quote outside braces does not swallow the following arguments.""" + mock_load_config.return_value = mock_configs + mock_path_exists.return_value = True + launcher = ModelLauncher("test-model", {}) + + args = launcher._process_engine_args("--x=it's,-tp=2", "vllm") + + assert args == {"--x": "it's", "--tensor-parallel-size": "2"} + + @pytest.mark.parametrize( + "arg_string", + [ + '-sc={"method":"ngram",-tp=2', + "-cc=[1,2,-tp=2", + '-sc={"method":"ngram},-tp=2', + "-tp=2}", + ], + ) + @patch("vec_inf.client._helper.utils.load_config") + @patch("pathlib.Path.exists") + def test_process_engine_args_unbalanced( + self, mock_path_exists, mock_load_config, arg_string, mock_configs + ): + """Unbalanced braces, brackets or quotes raise a clear error.""" + mock_load_config.return_value = mock_configs + mock_path_exists.return_value = True + launcher = ModelLauncher("test-model", {}) + + with pytest.raises(ValueError, match="engine arguments"): + launcher._process_engine_args(arg_string, "vllm") + @patch("vec_inf.client._helper.utils.load_config") @patch("pathlib.Path.exists") def test_process_engine_args_space_separated_values( @@ -1328,6 +1388,29 @@ def test_parse_sglang_per_rank_gauges(self, collector): assert parsed["kv_cache_usage"] == 0.5 assert parsed["finished_requests_total"] == 9.0 + def test_parse_sglang_dp_attention_ranks(self, collector): + """With --enable-dp-attention each DP group reports from its own rank. + + Here TP=4 is split into two DP attention groups; group 0 reports from + tp_rank 0 and group 1 from tp_rank 2. Both groups must be counted, and + a group that reports from several ranks must be counted once. + """ + text = """\ +sglang:num_running_reqs{dp_rank="0",model_name="m",pp_rank="0",tp_rank="0"} 5.0 +sglang:num_running_reqs{dp_rank="0",model_name="m",pp_rank="0",tp_rank="1"} 5.0 +sglang:num_running_reqs{dp_rank="1",model_name="m",pp_rank="0",tp_rank="2"} 3.0 +sglang:num_running_reqs{dp_rank="1",model_name="m",pp_rank="0",tp_rank="3"} 3.0 +sglang:num_queue_reqs{dp_rank="0",model_name="m",pp_rank="0",tp_rank="0"} 1.0 +sglang:num_queue_reqs{dp_rank="1",model_name="m",pp_rank="0",tp_rank="2"} 2.0 +sglang:token_usage{dp_rank="0",model_name="m",pp_rank="0",tp_rank="0"} 0.2 +sglang:token_usage{dp_rank="1",model_name="m",pp_rank="0",tp_rank="2"} 0.4 +""" + parsed = collector._parse_metrics(text) + + assert parsed["requests_running"] == 8.0 + assert parsed["requests_waiting"] == 3.0 + assert parsed["kv_cache_usage"] == pytest.approx(0.3) + def test_parse_vllm_data_parallel_engines(self, collector): """Samples from several engines are summed, KV cache usage is averaged.""" text = """\ diff --git a/tests/vec_inf/client/test_slurm_script_generator.py b/tests/vec_inf/client/test_slurm_script_generator.py index a03ff445..615b964b 100644 --- a/tests/vec_inf/client/test_slurm_script_generator.py +++ b/tests/vec_inf/client/test_slurm_script_generator.py @@ -10,6 +10,7 @@ from vec_inf.client._slurm_script_generator import ( BatchSlurmScriptGenerator, SlurmScriptGenerator, + format_engine_arg_value, ) @@ -262,11 +263,12 @@ def test_generate_launch_cmd_vllm_executor_backend( def test_engine_arg_values_survive_bash( self, basic_params, num_nodes, engine, tmp_path ): - """JSON and spaced values reach the engine exactly as configured. + """JSON, multi-value and quoted values reach the engine as intended. - The generated script must pass ``bash -n``, and the engine command, - run through bash with ``echo`` in place of the engine, must print each - value unchanged (no stripped quotes, no word splitting). + The generated script must pass ``bash -n``, and the engine arguments, + run through bash with ``printf`` in place of the engine, must come out + as the engine should see them: JSON as one word, flags that take + several values as several words, and a quoted path as one word. """ spec = '{"method":"ngram","num_speculative_tokens":3}' params = basic_params.copy() @@ -274,7 +276,8 @@ def test_engine_arg_values_survive_bash( params["num_nodes"] = num_nodes params["engine_args"] = { "--speculative-config": spec, - "--chat-template": "/path/with space/template.jinja", + "--chat-template": "'/path/with space/template.jinja'", + "--cudagraph-capture-sizes": "1 2 4 8", "--max-model-len": "8192", "--enforce-eager": True, } @@ -306,9 +309,42 @@ def test_engine_arg_values_survive_bash( echoed[echoed.index("--chat-template") + 1] == "/path/with space/template.jinja" ) + sizes = echoed.index("--cudagraph-capture-sizes") + assert echoed[sizes + 1 : sizes + 5] == ["1", "2", "4", "8"] assert echoed[echoed.index("--max-model-len") + 1] == "8192" assert "--enforce-eager" in echoed + @pytest.mark.parametrize( + ("value", "expected"), + [ + ("1 2 4 8", ["1", "2", "4", "8"]), + ("model-a model-b", ["model-a", "model-b"]), + ('{"a": "x $y", "b": [1, 2]}', ['{"a": "x $y", "b": [1, 2]}']), + ("[1, 2, 4]", ["[1, 2, 4]"]), + ("'/p q/t.jinja'", ["/p q/t.jinja"]), + ("price$5", ["price$5"]), + ("a\\$b", ["a$b"]), + ("$SLURM_TMPDIR/x", ["/tmp/job1/x"]), + ("${SLURM_TMPDIR}/x", ["/tmp/job1/x"]), + ("$SLURM_TMPDIR/a\\$b", ["/tmp/job1/a$b"]), + ("~/t.jinja", ["/home/u/t.jinja"]), + ("it's", ["it's"]), + ('q"x', ['q"x']), + ], + ) + def test_format_engine_arg_value_through_bash(self, value, expected): + """Each value expands in bash to exactly the expected words.""" + quoted = format_engine_arg_value(value) + words = subprocess.run( + ["bash", "-c", f"printf '%s\\n' {quoted}"], + env={"SLURM_TMPDIR": "/tmp/job1", "HOME": "/home/u", "y": "Y"}, + text=True, + capture_output=True, + check=True, + ).stdout.splitlines() + + assert words == expected + def test_engine_arg_values_keep_shell_expansion(self, basic_params): """Values with $VAR or a leading ~/ are still expanded by bash.""" params = basic_params.copy() @@ -335,13 +371,13 @@ def test_engine_arg_values_keep_shell_expansion(self, basic_params): @pytest.mark.parametrize("engine", ["vllm", "sglang"]) def test_server_port_leaves_room_for_sglang_grpc(self, basic_params, engine): - """SGLang binds server port + 10000 for gRPC, so cap the port at 55535.""" + """Server port is above Ray's worker ports and leaves room for SGLang gRPC.""" params = basic_params.copy() params["engine"] = engine setup = SlurmScriptGenerator(params)._generate_server_setup() assert ( - "server_port_number=$(find_available_port $head_node 8080 55535)" in setup + "server_port_number=$(find_available_port $head_node 20000 55535)" in setup ) def test_multinode_vllm_ray_port_outside_worker_range(self, multinode_params): @@ -363,6 +399,9 @@ def test_multinode_vllm_container_passes_nccl_env(self, singularity_params): assert all("$NCCL_ENV_ARG" in line for line in container_lines) # Unquoted so "--env NCCL_IB_GID_INDEX=3" splits into two arguments assert '"$NCCL_ENV_ARG"' not in content + # Without RDMA tools NCCL picks the transport itself (IB is not forced off) + assert 'export NCCL_ENV_ARG=""' in content + assert "NCCL_IB_DISABLE=1" not in content # The GID index default must be a parameter expansion, not literal text assert "${NCCL_IB_GID_INDEX:-3}" in content assert "${NCCL_IB_GID_INDEX:-0}" in content diff --git a/vec_inf/client/_client_vars.py b/vec_inf/client/_client_vars.py index 8b256566..f5e8fc39 100644 --- a/vec_inf/client/_client_vars.py +++ b/vec_inf/client/_client_vars.py @@ -82,10 +82,11 @@ # SGLang reports scheduler metrics once per tensor, pipeline and expert parallel # rank (with pipeline parallelism, or --enable-metrics-for-all-schedulers), and -# every rank sees the same requests. Only samples from rank 0 of these groups -# are counted, so running/waiting requests are not multiplied by the number of +# every rank of a data parallel group sees the same requests. For each metric and +# dp_rank only the samples from the lowest (pp_rank, tp_rank, moe_ep_rank) are +# counted, so running/waiting requests are not multiplied by the number of # ranks. Data parallel ranks (dp_rank) serve different requests and are summed. -SGLANG_PRIMARY_RANK_LABELS = ("tp_rank", "pp_rank", "moe_ep_rank") +SGLANG_PRIMARY_RANK_LABELS = ("pp_rank", "tp_rank", "moe_ep_rank") # Slurm job configuration arguments SLURM_JOB_CONFIG_ARGS = { diff --git a/vec_inf/client/_helper.py b/vec_inf/client/_helper.py index e2e08f9a..e82f3d1a 100644 --- a/vec_inf/client/_helper.py +++ b/vec_inf/client/_helper.py @@ -58,8 +58,11 @@ def _split_engine_args(arg_string: str) -> list[str]: """Split a comma-separated engine argument string. - Commas inside braces, brackets or quotes do not split, so JSON values such - as ``-sc={"method":"ngram","num_speculative_tokens":3}`` stay in one piece. + Commas inside braces or brackets (and inside quoted strings within them) + do not split, so JSON values such as + ``-sc={"method":"ngram","num_speculative_tokens":3}`` stay in one piece. + Quotes outside braces and brackets are ordinary characters, so a value + like ``it's`` does not swallow the arguments after it. Parameters ---------- @@ -70,6 +73,11 @@ def _split_engine_args(arg_string: str) -> list[str]: ------- list[str] The individual arguments + + Raises + ------ + ValueError + If a brace, bracket or quoted string inside one is left open """ parts: list[str] = [] current: list[str] = [] @@ -79,17 +87,25 @@ def _split_engine_args(arg_string: str) -> list[str]: if quote: if char == quote: quote = "" - elif char in "\"'": + elif depth > 0 and char in "\"'": quote = char elif char in "{[": depth += 1 elif char in "}]": - depth = max(depth - 1, 0) + if depth == 0: + raise ValueError( + f"Unbalanced '{char}' in engine arguments: {arg_string!r}" + ) + depth -= 1 elif char == "," and depth == 0: parts.append("".join(current)) current = [] continue current.append(char) + if quote: + raise ValueError(f"Unclosed {quote} quote in engine arguments: {arg_string!r}") + if depth: + raise ValueError(f"Unclosed '{{' or '[' in engine arguments: {arg_string!r}") parts.append("".join(current)) return parts @@ -1069,6 +1085,52 @@ def _parse_labels(label_text: str) -> dict[str, str]: for name, value in _PROM_LABEL_RE.findall(label_text) } + @staticmethod + def _primary_rank_samples( + samples: list[tuple[str, dict[str, str], float]], + ) -> list[tuple[str, float]]: + """Keep one reporting rank per SGLang data parallel group. + + SGLang repeats scheduler metrics for every tensor, pipeline and expert + parallel rank of a data parallel group (with pipeline parallelism, or + ``--enable-metrics-for-all-schedulers``). Within each metric and + ``dp_rank`` only the samples from the lowest rank are kept, so the + group is counted once whichever rank reports it (with + ``--enable-dp-attention`` the reporting rank need not be tp_rank 0). + Different ``dp_rank`` values serve different requests and are kept. + + Parameters + ---------- + samples : list[tuple[str, dict[str, str], float]] + (metric name, labels, value) for each SGLang sample + + Returns + ------- + list[tuple[str, float]] + (metric name, value) for the samples to aggregate + """ + + def rank_key(labels: dict[str, str]) -> tuple[int, ...]: + ranks = [] + for label in SGLANG_PRIMARY_RANK_LABELS: + try: + ranks.append(int(labels.get(label, "0"))) + except ValueError: + ranks.append(0) + return tuple(ranks) + + lowest: dict[tuple[str, str], tuple[int, ...]] = {} + for metric_name, labels, _ in samples: + group = (metric_name, labels.get("dp_rank", "")) + key = rank_key(labels) + if group not in lowest or key < lowest[group]: + lowest[group] = key + return [ + (metric_name, value) + for metric_name, labels, value in samples + if rank_key(labels) == lowest[(metric_name, labels.get("dp_rank", ""))] + ] + def _parse_metrics(self, metrics_text: str) -> dict[str, float]: """Parse Prometheus text exposition into vec-inf metric values. @@ -1088,6 +1150,8 @@ def _parse_metrics(self, metrics_text: str) -> dict[str, float]: Parsed metrics as key-value pairs """ samples: dict[str, list[float]] = {} + # SGLang samples are kept with their rank labels and filtered below + sglang_samples: list[tuple[str, dict[str, str], float]] = [] for raw_line in metrics_text.splitlines(): line = raw_line.strip() if not line or line.startswith("#"): @@ -1106,15 +1170,17 @@ def _parse_metrics(self, metrics_text: str) -> dict[str, float]: labels.get(k) in values for k, values in excluded.items() ): continue - if metric_name.startswith("sglang:") and any( - labels.get(k, "0") != "0" for k in SGLANG_PRIMARY_RANK_LABELS - ): - continue try: value = float(value_text) except ValueError: continue + if metric_name.startswith("sglang:"): + sglang_samples.append((metric_name, labels, value)) + else: + samples.setdefault(metric_name, []).append(value) + + for metric_name, value in self._primary_rank_samples(sglang_samples): samples.setdefault(metric_name, []).append(value) parsed: dict[str, float] = {} diff --git a/vec_inf/client/_slurm_script_generator.py b/vec_inf/client/_slurm_script_generator.py index 568547dc..277a7cdf 100644 --- a/vec_inf/client/_slurm_script_generator.py +++ b/vec_inf/client/_slurm_script_generator.py @@ -4,6 +4,7 @@ servers in both single-node and multi-node configurations. """ +import re import shlex from datetime import datetime from pathlib import Path @@ -18,15 +19,47 @@ from vec_inf.client._slurm_vars import CONTAINER_MODULE_NAME, IMAGE_PATH +# A shell variable reference that is not escaped: $NAME or ${...} +_SHELL_VAR_RE = re.compile(r"(? str: + """Quote one shell word, keeping real variable references expandable. + + Parameters + ---------- + token : str + One word of an engine argument value + + Returns + ------- + str + The word, quoted for bash + """ + if _SHELL_VAR_RE.search(token) or "\\$" in token: + # Double quotes keep $VAR and ${VAR} expanding. Backslashes are left + # as they are, so an escaped \$ stays a literal dollar sign, as it + # would on a command line. + return '"' + token.replace('"', '\\"').replace("`", "\\`") + '"' + if token.startswith("~/"): + return "~/" + shlex.quote(token[2:]) + return shlex.quote(token) + + def format_engine_arg_value(value: Any) -> str: r"""Quote an engine argument value for the generated bash script. - Values are single-quoted with ``shlex.quote`` so JSON (e.g. for - ``--speculative-config``) and values with spaces reach the engine - unchanged. Two cases keep shell expansion on purpose: a value containing - ``$`` is double-quoted (escaping ``\\``, ``"`` and backticks) so variables - such as ``$SLURM_TMPDIR`` still expand, and a leading ``~/`` is left - unquoted so it still expands to the home directory. + * JSON values (starting with ``{`` or ``[``) are single-quoted as one + word, so they reach the engine unchanged. + * Other values are split into words with ``shlex.split`` and each word is + quoted on its own, so flags that take several values keep working + (``--cudagraph-capture-sizes 1 2 4 8``, ``--served-model-name a b``). + Quote a value that is one word with spaces in it, for example + ``--chat-template='/path/with space/t.jinja'``. + * A word that references a shell variable (``$NAME`` or ``${...}``) is + double-quoted so the variable still expands, and a leading ``~/`` is + left unquoted so it still expands to the home directory. Any other + ``$`` stays literal. Parameters ---------- @@ -39,12 +72,22 @@ def format_engine_arg_value(value: Any) -> str: The value, quoted for bash """ text = str(value) - if "$" in text: - escaped = text.replace("\\", "\\\\").replace('"', '\\"').replace("`", "\\`") - return f'"{escaped}"' - if text.startswith("~/"): - return "~/" + shlex.quote(text[2:]) - return shlex.quote(text) + stripped = text.strip() + if stripped.startswith(("{", "[")): + return shlex.quote(stripped) + # Split like a shell would on whitespace and quotes, but keep backslashes + # so escaped dollar signs survive to _quote_token + lexer = shlex.shlex(text, posix=True) + lexer.whitespace_split = True + lexer.escape = "" + try: + tokens = list(lexer) + except ValueError: + # Unbalanced quotes: pass the value through as a single word + tokens = [text] + if not tokens: + return shlex.quote(text) + return " ".join(_quote_token(token) for token in tokens) class SlurmScriptGenerator: diff --git a/vec_inf/client/_slurm_templates.py b/vec_inf/client/_slurm_templates.py index d34a8afd..e638141a 100644 --- a/vec_inf/client/_slurm_templates.py +++ b/vec_inf/client/_slurm_templates.py @@ -144,9 +144,10 @@ class SlurmScriptTemplate(TypedDict): 'head_node_ip=$(srun --nodes=1 --ntasks=1 -w "$head_node" hostname --ip-address)', "\n# Check for RDMA devices and set environment variable accordingly", "if ! command -v ibv_devices >/dev/null 2>&1; then", - ' echo "ibv_devices not found; forcing TCP. (No RDMA userland on host?)"', - " export NCCL_IB_DISABLE=1", - ' export NCCL_ENV_ARG="--env NCCL_IB_DISABLE=1"', + " # No RDMA tools on the host: leave the choice to NCCL, which", + " # falls back to TCP sockets when it finds no usable IB device.", + ' echo "ibv_devices not found; letting NCCL pick the transport"', + ' export NCCL_ENV_ARG=""', "else", " # Pick GID index based on link layer (IB vs RoCE)", ' if ibv_devinfo 2>/dev/null | grep -q "link_layer:.*Ethernet"; then', @@ -195,9 +196,10 @@ class SlurmScriptTemplate(TypedDict): ], }, "find_server_port": [ - # Max 55535: SGLang also binds a gRPC port at server port + 10000 and - # refuses to start if that is above 65535 - "\nserver_port_number=$(find_available_port $head_node 8080 55535)", + # 20000-55535: stay above Ray's worker port range (10002-19999), and + # SGLang also binds a gRPC port at server port + 10000, which must not + # go above 65535 + "\nserver_port_number=$(find_available_port $head_node 20000 55535)", 'server_address="http://${head_node}:${server_port_number}/v1"', ], "write_to_json": [ @@ -320,8 +322,9 @@ class BatchModelLaunchScriptTemplate(TypedDict): "server_address_setup": [ "source {src_dir}/find_port.sh", "head_node_ip=${{SLURMD_NODENAME}}", - # Max 55535: SGLang also binds a gRPC port at server port + 10000 - "server_port_number=$(find_available_port $head_node_ip 8080 55535)", + # 20000-55535: above Ray's worker port range, and SGLang also binds a + # gRPC port at server port + 10000 + "server_port_number=$(find_available_port $head_node_ip 20000 55535)", 'server_address="http://${{head_node_ip}}:${{server_port_number}}/v1"\n', "echo $server_address\n", ], From e05225aaac0dc93615c30d9b9d214922aae2dacf Mon Sep 17 00:00:00 2001 From: Amrit Krishnan Date: Sun, 25 Oct 2026 16:13:02 -0400 Subject: [PATCH 10/10] Third review: env order, double-quote escaping, $$ and apostrophes - Put $NCCL_ENV_ARG before the user's --env values in the multi-node container command; apptainer keeps the last value for a repeated key, so --env NCCL_IB_GID_INDEX=2 now overrides the script default. - Inside double quotes, escape every backslash except one directly before $, so a value like $X\"q no longer leaves an open quote. - $$ (and $$NAME) no longer counts as a variable reference. - Split values into words on whitespace with quotes only grouping words when they start a word; apostrophes inside a word (it's, a'b'c) stay literal instead of being stripped as a quote pair. - Docs: override note for NCCL_IB_GID_INDEX and the quoting rules. --- docs/user_guide.md | 4 +- .../client/test_slurm_script_generator.py | 24 ++++ vec_inf/client/_slurm_script_generator.py | 106 ++++++++++++++---- vec_inf/client/_slurm_templates.py | 4 +- 4 files changed, 116 insertions(+), 22 deletions(-) diff --git a/docs/user_guide.md b/docs/user_guide.md index f1f3eb0b..b21d3c96 100644 --- a/docs/user_guide.md +++ b/docs/user_guide.md @@ -416,8 +416,8 @@ The vLLM image moved from vLLM 0.22.0 to 0.32.0. Things you may notice: * Commas inside `{}` or `[]` (and inside quoted strings within them) no longer split arguments, so JSON values such as `--vllm-args '-sc={"method":"ngram","num_speculative_tokens":3}'` reach the engine unchanged. Unbalanced braces, brackets or quotes are reported as an error. * A flag and its value can be separated by `=` or a space: `-tp=2` and `-tp 2` are the same. * Values are quoted in the generated Slurm script. A value starting with `{` or `[` is passed as one word. Other values are split into words, so flags that take several values work in both forms, for example `--cudagraph-capture-sizes=1 2 4 8` or `--served-model-name a b`. To pass one word that contains spaces, quote it: `--chat-template='/path/with space/t.jinja'`. - * A word that references a variable (`$NAME` or `${NAME}`) is still expanded by the shell, `\$` gives a literal dollar sign, and a leading `~/` still expands to your home directory. Any other `$` (for example `price$5`) is passed literally. -* **Multi-node networking**: multi-node vLLM containers now receive the NCCL settings from the launch script's RDMA check. On nodes with RDMA tools, RoCE links get `NCCL_IB_GID_INDEX=3` and InfiniBand links `0` (override with `--env NCCL_IB_GID_INDEX=...`). Without `ibv_devices` on the host the script no longer forces `NCCL_IB_DISABLE=1`; NCCL picks the transport itself and falls back to TCP sockets when there is no usable device. + * A word that references a variable (`$NAME` or `${NAME}`) is still expanded by the shell, `\$` gives a literal dollar sign, and a leading `~/` still expands to your home directory. Any other `$` (for example `price$5` or `$$`) is passed literally. Quotes only group words when they start a word; apostrophes inside a word (`it's`) are kept as they are. +* **Multi-node networking**: multi-node vLLM containers now receive the NCCL settings from the launch script's RDMA check. On nodes with RDMA tools, RoCE links get `NCCL_IB_GID_INDEX=3` and InfiniBand links `0`. To use another index, pass it at launch, for example `vec-inf launch ... --env NCCL_IB_GID_INDEX=2`; your `--env` values come after the script's defaults in the container command, so they take precedence. Without `ibv_devices` on the host the script no longer forces `NCCL_IB_DISABLE=1`; NCCL picks the transport itself and falls back to TCP sockets when there is no usable device. * **Ports**: the server port is now picked between 20000 and 55535, and the Ray head port between 20000 and 65535. ### Supported GPUs diff --git a/tests/vec_inf/client/test_slurm_script_generator.py b/tests/vec_inf/client/test_slurm_script_generator.py index 615b964b..10fad8b8 100644 --- a/tests/vec_inf/client/test_slurm_script_generator.py +++ b/tests/vec_inf/client/test_slurm_script_generator.py @@ -330,6 +330,16 @@ def test_engine_arg_values_survive_bash( ("~/t.jinja", ["/home/u/t.jinja"]), ("it's", ["it's"]), ('q"x', ['q"x']), + # Backslashes inside double quotes are escaped, except before $ + ('$SLURM_TMPDIR\\"q', ['/tmp/job1\\"q']), + ("$SLURM_TMPDIR\\\\x", ["/tmp/job1\\\\x"]), + # $$ is not a variable reference + ("$$", ["$$"]), + ("$$SLURM_TMPDIR", ["$$SLURM_TMPDIR"]), + # Quotes inside a word are literal, a leading quote groups words + ("a'b'c", ["a'b'c"]), + ("don't stop", ["don't", "stop"]), + ('"/p q/t" x', ["/p q/t", "x"]), ], ) def test_format_engine_arg_value_through_bash(self, value, expected): @@ -407,6 +417,20 @@ def test_multinode_vllm_container_passes_nccl_env(self, singularity_params): assert "${NCCL_IB_GID_INDEX:-0}" in content assert "={NCCL_IB_GID_INDEX" not in content + def test_multinode_user_env_overrides_nccl_defaults(self, singularity_params): + """User --env values come after $NCCL_ENV_ARG so they win in apptainer.""" + multinode = singularity_params.copy() + multinode["num_nodes"] = "2" + multinode["env"] = {"NCCL_IB_GID_INDEX": "2"} + content = SlurmScriptGenerator(multinode)._generate_script_content() + + container_lines = [ + line for line in content.splitlines() if "--contain --cleanenv" in line + ] + assert len(container_lines) == 3 + for line in container_lines: + assert line.index("$NCCL_ENV_ARG") < line.index("--env NCCL_IB_GID_INDEX=2") + def test_multinode_vllm_container_shares_pid_namespace(self, singularity_params): """Ray and vllm serve containers must share the host PID namespace. diff --git a/vec_inf/client/_slurm_script_generator.py b/vec_inf/client/_slurm_script_generator.py index 277a7cdf..01a46183 100644 --- a/vec_inf/client/_slurm_script_generator.py +++ b/vec_inf/client/_slurm_script_generator.py @@ -19,8 +19,80 @@ from vec_inf.client._slurm_vars import CONTAINER_MODULE_NAME, IMAGE_PATH -# A shell variable reference that is not escaped: $NAME or ${...} -_SHELL_VAR_RE = re.compile(r"(? list[str]: + """Split a value into words on whitespace. + + A word that starts with a single or double quote runs to the matching + closing quote and may contain spaces; the quotes are removed. Quotes + anywhere else are ordinary characters, so apostrophes inside a word + (``it's``, ``a'b'c``) are kept. Backslashes are kept as they are. + + Parameters + ---------- + text : str + The value to split + + Returns + ------- + list[str] + The words + + Raises + ------ + ValueError + If a word starts with a quote that is never closed + """ + words: list[str] = [] + i, n = 0, len(text) + while i < n: + if text[i].isspace(): + i += 1 + continue + if text[i] in "'\"": + end = text.find(text[i], i + 1) + if end == -1: + raise ValueError(f"Unclosed {text[i]} quote in {text!r}") + word = text[i + 1 : end] + i = end + 1 + else: + word = "" + start = i + while i < n and not text[i].isspace(): + i += 1 + words.append(word + text[start:i]) + return words + + +def _double_quote(token: str) -> str: + """Double-quote a word so variables expand but nothing else does. + + Every backslash is escaped except one directly before ``$``, which keeps + escaping that dollar sign. ``"`` and backticks are escaped. + + Parameters + ---------- + token : str + The word + + Returns + ------- + str + The word in double quotes + """ + out = [] + for i, char in enumerate(token): + if char == "\\": + out.append("\\" if token[i + 1 : i + 2] == "$" else "\\\\") + elif char in '"`': + out.append("\\" + char) + else: + out.append(char) + return '"' + "".join(out) + '"' def _quote_token(token: str) -> str: @@ -37,10 +109,9 @@ def _quote_token(token: str) -> str: The word, quoted for bash """ if _SHELL_VAR_RE.search(token) or "\\$" in token: - # Double quotes keep $VAR and ${VAR} expanding. Backslashes are left - # as they are, so an escaped \$ stays a literal dollar sign, as it - # would on a command line. - return '"' + token.replace('"', '\\"').replace("`", "\\`") + '"' + # Double quotes keep $VAR and ${VAR} expanding, and an escaped \$ + # stays a literal dollar sign, as it would on a command line. + return _double_quote(token) if token.startswith("~/"): return "~/" + shlex.quote(token[2:]) return shlex.quote(token) @@ -51,15 +122,17 @@ def format_engine_arg_value(value: Any) -> str: * JSON values (starting with ``{`` or ``[``) are single-quoted as one word, so they reach the engine unchanged. - * Other values are split into words with ``shlex.split`` and each word is - quoted on its own, so flags that take several values keep working + * Other values are split into words on whitespace and each word is quoted + on its own, so flags that take several values keep working (``--cudagraph-capture-sizes 1 2 4 8``, ``--served-model-name a b``). - Quote a value that is one word with spaces in it, for example - ``--chat-template='/path/with space/t.jinja'``. + A word that starts with a quote runs to the matching quote, so a value + with spaces can be passed as one word: + ``--chat-template='/path/with space/t.jinja'``. Quotes inside a word + are kept literally. * A word that references a shell variable (``$NAME`` or ``${...}``) is - double-quoted so the variable still expands, and a leading ``~/`` is - left unquoted so it still expands to the home directory. Any other - ``$`` stays literal. + double-quoted so the variable still expands, ``\$`` gives a literal + dollar sign, and a leading ``~/`` is left unquoted so it still expands + to the home directory. Any other ``$`` (including ``$$``) stays literal. Parameters ---------- @@ -75,13 +148,8 @@ def format_engine_arg_value(value: Any) -> str: stripped = text.strip() if stripped.startswith(("{", "[")): return shlex.quote(stripped) - # Split like a shell would on whitespace and quotes, but keep backslashes - # so escaped dollar signs survive to _quote_token - lexer = shlex.shlex(text, posix=True) - lexer.whitespace_split = True - lexer.escape = "" try: - tokens = list(lexer) + tokens = _split_words(text) except ValueError: # Unbalanced quotes: pass the value through as a single word tokens = [text] diff --git a/vec_inf/client/_slurm_templates.py b/vec_inf/client/_slurm_templates.py index e638141a..7777784e 100644 --- a/vec_inf/client/_slurm_templates.py +++ b/vec_inf/client/_slurm_templates.py @@ -129,7 +129,9 @@ class SlurmScriptTemplate(TypedDict): # $NCCL_ENV_ARG (set by the RDMA check in multinode_vllm, e.g. # "--env NCCL_IB_GID_INDEX=3") is left unquoted on purpose so it splits into # apptainer arguments; --cleanenv would otherwise drop the NCCL_IB_* exports. - "container_command_ray": f"{CONTAINER_MODULE_NAME} exec --nv {{env_str}} $NCCL_ENV_ARG --contain --cleanenv {{image_path}} \\", + # It comes before {env_str}: for repeated --env keys apptainer keeps the + # last value, so a user's --env NCCL_IB_GID_INDEX=... wins. + "container_command_ray": f"{CONTAINER_MODULE_NAME} exec --nv $NCCL_ENV_ARG {{env_str}} --contain --cleanenv {{image_path}} \\", "activate_venv": "source {venv}/bin/activate", "server_setup": { "single_node": [