Skip to content

Qwen3.5-397B RL trainer on 128 v7x: pad-first sequence packing, fused sharded log-prob head VJP, and the tuned trainer recipe - #2913

Open
parambole wants to merge 4 commits into
atwigg/mlperffrom
parambole/rl-trainer-4600
Open

Qwen3.5-397B RL trainer on 128 v7x: pad-first sequence packing, fused sharded log-prob head VJP, and the tuned trainer recipe#2913
parambole wants to merge 4 commits into
atwigg/mlperffrom
parambole/rl-trainer-4600

recipes(397b v7x): TRAINER_TUNED_FLAGS covers the 256-chip trainer me…

8c75ae3
Select commit
Loading
Failed to load commit list.

Select a check to view from the sidebar