All pull requests

Pull requests

• Pre-pull static R2E-Gym sandbox container images and drop OS caches before init_start

  #2816·andytwigg opened Oct 11, 2026Collaborator·

  ·
  13

  

• Non-blocking sandbox wait_initial and episode timer reset fix

  #2815·andytwigg opened Oct 11, 2026Collaborator·

  ·
  3

  

• Ahead-of-time trainer compilation and graph warmup during init

  #2814·andytwigg opened Oct 11, 2026Collaborator·

  ·
  13

  

• Update default flags in mlperf_gbs1024_submission.sh

  #2813·andytwigg opened Oct 11, 2026Collaborator·6 tasks·

  ·
  2

  

• deepswe: step-0 knobs INITIAL_DISPATCH_STAGGER_S and TRAINER_WARMUP_COMPILE (stacks on #2801)

  #2812·ultrons opened Oct 11, 2026·

  ·
  

• [Recipes][MoE] Plumb MoE optimization switches across launcher and recipes

  #2811·wenxindongwork opened Oct 10, 2026Collaborator·

  ·
  

• Skip the token after left padding in the ORPO prompt loss

  #2809·vineethsaivs opened Oct 10, 2026·6 tasks done·

  ·
  5

  

• [recipes] Key the rollout JAX compilation cache on USE_BATCHED_RPA_LONG_CTX_KERNEL

  #2807·wenxindongwork opened Oct 10, 2026Collaborator·

  ·
  4

  

• recipes: add RAGGED_GATHER_REDUCE_VERSION switch (default v3) + bump tpu-inference pin

  #2804·wenxindongwork opened Oct 10, 2026Collaborator·

  ·
  10

  

• Retry vLLM engine start in the rollout sampler

  #2803·ultrons opened Oct 10, 2026·

  ·
  3

  

• fix(orchestrator): resolve remote_worker_handles() attribute error after PR #2767 merge

  #2802·lc5211 opened Oct 10, 2026Collaborator·

  ·
  7

  

• feat(deepswe): SANDBOX_WAIT_INITIAL knob to skip the start-up warm-pool readiness barrier

  #2801·ultrons opened Oct 10, 2026·

  ·
  

• Seed each eval rollout engine from --seed and replica index

  #2795·susanbao opened Oct 10, 2026Collaborator·

  ·
  3

  

• fix(openhands): bake chardet into agent server and pre-seed site-packages

  #2792·niting opened Oct 9, 2026Collaborator·

  ·
  5

  

• This CL:

  #2791·copybara-service[bot] opened Oct 9, 2026Bot·

  ·
  1

  

• Precompile trainer fwd_bwd, update, and eval steps during configure_worker

  #2789·lc5211 opened Oct 9, 2026Collaborator·

  ·
  6

  

• sft: keep custom metadata in the final checkpoint saved by close()

  #2788·yangziao56 opened Oct 9, 2026·6 tasks done·

  ·
  4

  

• agentic: stop the rollout producer when train() raises

  #2786·yangziao56 opened Oct 9, 2026·6 tasks done·

  ·
  4

  

• [Tunix] Persist first-turn prompt token ids and masked_out in the trajectory store.

  #2777·copybara-service[bot] opened Oct 9, 2026Bot·

  ·
  

• fix(recipes): update v5p 35B/397B recipes and multi-host jobset templ…

  #2775·knightwh opened Oct 9, 2026Collaborator·6 tasks·

  ·
  5

  

• fix(deployment): make colocated-python-sidecar honor SIGTERM; bound trainer teardown

  #2772·YixuanWang-99 opened Oct 8, 2026Collaborator·

  ·
  5

  

• cli: honor optimizer_config.max_grad_norm in create_optimizer

  #2758·yangziao56 opened Oct 8, 2026·6 tasks done·

  ·
  4

  

• Add the two inspection subcommands of trajectory-explore, both built on the shared commands/stats.py derivation so they report identical numbers:

  #2757·copybara-service[bot] opened Oct 8, 2026Bot·

  ·
  

• FileTrajectoryStore.get_trajectories and get_trajectories_metadata accessed one file at a time, so on remote filesystems such as GCS every existence check, read, and directory listing added its full latency to the call.

  #2756·copybara-service[bot] opened Oct 8, 2026Bot·

  ·
  1

  

• Capturing an agent step costs the rollout thread time proportional to its token count, in three places. converter.create_agent_step converted token IDs and logprobs one NumPy scalar at a time. Pydantic validated each IntArray and FloatArray field by first converting the array to a list, which it discarded once the array matched np.ndarray exactly. WriteTask and InMemoryTrajectoryStore deep-copy every step and metadata with model_copy(deep=True), which copies token and logprob lists one element at a time.

  #2754·copybara-service[bot] opened Oct 8, 2026Bot·

  ·
  1