Skip to content

[fsdp, megatron] fix: Engine Rollout Worker LoRA Parameter Update #4836

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
wuxibin89 merged 3 commits into from
Jan 9, 2026
Merged

[fsdp, megatron] fix: Engine Rollout Worker LoRA Parameter Update #4836

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
5692b9e
Layered summon
JacobHelwig Jan 8, 2026
86c7998
Fix megatron bridge
JacobHelwig Jan 8, 2026
3b581eb
Use layered summon attr
JacobHelwig Jan 8, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 5 additions & 2 deletions verl/workers/engine/megatron/transformer_impl.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -536,9 +536,12 @@ def forward_backward_batch(self, data: TensorDict, loss_function: Callable, forw
else:
return {}

def get_per_tensor_param(self):
def get_per_tensor_param(self, **kwargs):
load_megatron_model_to_gpu(self.module, load_grad=False)
per_tensor_param = self.bridge.export_weights(self.module)
if self.vanilla_bridge:
per_tensor_param = self.bridge.export_weights(self.module)
else:
per_tensor_param = self.bridge.export_hf_weights(self.module)
# TODO: support megatron LoRA
return per_tensor_param, None

Expand Down
4 changes: 3 additions & 1 deletion verl/workers/engine_workers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -563,7 +563,9 @@ async def wake_up(self):
set_expandable_segments(False)

# 1. get per tensor generator from engine, this will load model to gpu
per_tensor_param, peft_config = self.actor.engine.get_per_tensor_param()
per_tensor_param, peft_config = self.actor.engine.get_per_tensor_param(
Copy link
Collaborator

@wuxibin89 wuxibin89 Jan 8, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@JacobHelwig Megatron engine get_per_tensor_param doesn't accept layered_summon and base_sync_done, please fix it.
https://github.com/volcengine/verl/blob/main/verl/workers/engine/megatron/transformer_impl.py#L539

Copy link
Contributor Author

@JacobHelwig JacobHelwig Jan 8, 2026 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good catch, thank you. I fixed it using kwargs in the Megatron engine and added a fix for LoRA with the Megatron engine (please see updated description).

layered_summon=self.layered_summon, base_sync_done=self.base_sync_done
)

# 2. resume weights and update weights
if self.config.rollout.free_cache_engine:
Expand Down