Skip to content

[Bugfix] fix Qwen3VLMoe load when pp > 1 #25838

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
merged 1 commit into from
Sep 28, 2025
Merged

[Bugfix] fix Qwen3VLMoe load when pp > 1 #25838

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
4 changes: 2 additions & 2 deletions vllm/model_executor/models/qwen3_vl_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -212,6 +212,8 @@ def load_weights(self, weights: Iterable[tuple[str,
# attempted to load as other weights later
is_expert_weight = True
name_mapped = name.replace(weight_name, param_name)
if is_pp_missing_parameter(name_mapped, self):
continue
if is_fused_expert:
loaded_weight = loaded_weight.transpose(-1,
-2) # no bias
Expand All @@ -230,8 +232,6 @@ def load_weights(self, weights: Iterable[tuple[str,
name_mapped, params_dict, loaded_weight,
shard_id, num_experts)
else:
if is_pp_missing_parameter(name_mapped, self):
continue
# Skip loading extra parameters for GPTQ/modelopt models
if name_mapped.endswith(
ignore_suffixes
Expand Down