Skip to content

[None][fix] Update the attention layers counting for Qwen3-next. #9072

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
Nov 16, 2025
Merged

[None][fix] Update the attention layers counting for Qwen3-next. #9072

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
7 changes: 7 additions & 0 deletions tensorrt_llm/_torch/model_config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -605,5 +605,12 @@ def get_layer_types(self) -> Optional[List[LayerTypeCpp]]:
def get_num_attention_layers(self):
if is_nemotron_hybrid(self.pretrained_config):
return self.pretrained_config.hybrid_override_pattern.count("*")
elif hasattr(
self.pretrained_config, "architectures"
) and self.pretrained_config.architectures is not None and self.pretrained_config.architectures[
0] in ["Qwen3NextForCausalLM"]:
# Qwen3NextForCausalLM has hybrid attention pattern(1:3 full attention:linear attention),
# we need to calculate the number of fullattention layers
return self.pretrained_config.num_hidden_layers // self.pretrained_config.full_attention_interval
else:
return self.pretrained_config.num_hidden_layers