Skip to content

[fsdp, model]{fix} Allow flash attention 2 to be used for NemotronH model on FSDP #4566

[fsdp, model]{fix} Allow flash attention 2 to be used for NemotronH model on FSDP

[fsdp, model]{fix} Allow flash attention 2 to be used for NemotronH model on FSDP #4566

This workflow is awaiting approval from a maintainer in #5419
Triggered via pull request February 26, 2026 20:29
Status Action required
Total duration
Artifacts
This workflow is awaiting approval from a maintainer in #5419
setup
setup
e2e_ppo_trainer_fsdp-qwen2_5vl-3b
e2e_ppo_trainer_fsdp-qwen2_5vl-3b
e2e_ppo_trainer_fsdp_vllm
e2e_ppo_trainer_fsdp_vllm
e2e_ppo_trainer_megatron-moe-expert-parallel
e2e_ppo_trainer_megatron-moe-expert-parallel
cleanup
cleanup
Fit to window
Zoom out
Zoom in