Skip to content

[bug] Switch to reload_weights API for loading weights in legacy inference codepath #1685

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
erictang000 merged 1 commit into from
Jun 1, 2026
Merged

[bug] Switch to reload_weights API for loading weights in legacy inference codepath #1685

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
5 changes: 4 additions & 1 deletion skyrl/backends/skyrl_train/inference_servers/vllm_worker.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,6 +82,8 @@ def load_weights(self, request: bytes) -> None:
"""
import pickle

from vllm.config import set_current_vllm_config

# Unpickle request to restore the original object type
assert isinstance(request, bytes), f"Expected bytes, got {type(request).__name__}"
request = pickle.loads(request)
Expand All @@ -90,7 +92,8 @@ def load_weights(self, request: bytes) -> None:
for name, tensor in self._weight_receiver.receive_weights(request):
weight_list.append((name, tensor))

self.model_runner.model.load_weights(weights=weight_list)
with torch.device(self.device), set_current_vllm_config(self.vllm_config):
self.model_runner.reload_weights(weights_iterator=iter(weight_list))
Comment thread
erictang000 marked this conversation as resolved.

for weight in weight_list:
del weight
Expand Down
Loading