Skip to content

why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? #28

why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice?

why are there multiple settings for actor_rollout_ref.model.enable_gradient_checkpointing? Is this a deliberate design choice? #28

This workflow is awaiting approval from a maintainer in #4263
Triggered via pull request February 5, 2026 07:01
@khazickhazic
synchronize #4263
khazic:main
Status Action required
Total duration
Artifacts
This workflow is awaiting approval from a maintainer in #4263

e2e_transferqueue_ascend.yml

on: pull_request
e2e_transferqueue_fsdp_ascend
e2e_transferqueue_fsdp_ascend
e2e_transferqueue_megatron_ascend
e2e_transferqueue_megatron_ascend
Fit to window
Zoom out
Zoom in