Skip to content

Updated reduce sum calculation to use einsum for gpt_oss #754

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
ochougul merged 2 commits into from
Jan 27, 2026
Merged

Updated reduce sum calculation to use einsum for gpt_oss #754

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
6f6ddc2
Updated reduce sum calculation to use einsum
asmigosw Jan 23, 2026
b164735
Ruff format
asmigosw Jan 23, 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
5 changes: 2 additions & 3 deletions QEfficient/transformers/models/gpt_oss/modeling_gpt_oss.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -402,9 +402,8 @@ def forward(self, hidden_states):

# Apply routing weights AFTER expert computation
experts_out = experts_out * router_top_value.unsqueeze(-1)
experts_out = experts_out.sum(dim=1)

return experts_out, router_logits
experts_out_sum = torch.einsum("bnd->bd", experts_out)
return experts_out_sum, router_logits

def optimized_moe_forward(self, hidden_states: torch.Tensor):
B, S, H = hidden_states.shape
Expand Down
Loading