Skip to content

refactor: remove gate normalization in DecoderUnit #222

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
hzhangxyz merged 1 commit into from
Mar 27, 2026
Merged

refactor: remove gate normalization in DecoderUnit #222

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: 2 additions & 3 deletions qmp/networks/transformers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -143,9 +143,8 @@ def forward(
similarity = torch.nn.functional.softmax(x @ self.centroid.t(), dim=-1)
# top_k_indices: batch * site * selected
_, top_k_indices = torch.topk(similarity + self.bias, self.selected_num, dim=-1)
# gate_prime, gate: batch * site * routed
gate_prime = torch.zeros_like(similarity).scatter_(-1, top_k_indices, similarity.gather(-1, top_k_indices))
gate = gate_prime / gate_prime.sum(dim=-1).unsqueeze(-1)
# gate, gate: batch * site * routed

Copilot AI Mar 27, 2026

Copy link

Choose a reason for hiding this comment

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

The inline comment is now inaccurate/typoed after removing gate_prime (it reads gate, gate: ...). Please update it to describe the single gate tensor to avoid confusion when reading the routing logic.

Suggested change
# gate, gate: batch * site * routed
# gate: batch * site * routed (top-k routed gating weights)

Copilot uses AI. Check for mistakes.
gate = torch.zeros_like(similarity).scatter_(-1, top_k_indices, similarity.gather(-1, top_k_indices))
for i, expert in enumerate(self.feed_forward_routed):
y = y + expert(x) * gate[:, :, i].unsqueeze(-1)
x = self.norm2(y)
Expand Down
Loading