Skip to content

Use one beta per head in the Gated DeltaNet code #1119

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

Open
Medamineelkhattabi wants to merge 1 commit into
base: main
Choose a base branch
from
+2 −2
Open

Use one beta per head in the Gated DeltaNet code #1119

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 ch04/08_deltanet/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -162,7 +162,7 @@ class GatedDeltaNet(nn.Module):
####################################################
### NEW: Gates for delta rule and output gating
self.W_gate = nn.Linear(d_in, d_out, bias=False)
self.W_beta = nn.Linear(d_in, d_out, bias=False)
self.W_beta = nn.Linear(d_in, num_heads, bias=False) # one beta per head

# Note: The decay gate alpha corresponds to
# A_log + W_alpha(x) + dt_bias
Expand Down Expand Up @@ -199,7 +199,7 @@ class GatedDeltaNet(nn.Module):
keys = keys.view(b, num_tokens, self.num_heads, self.head_dim)
values = values.view(b, num_tokens, self.num_heads, self.head_dim)
queries = queries.view(b, num_tokens, self.num_heads, self.head_dim)
beta = beta.view(b, num_tokens, self.num_heads, self.head_dim)
beta = beta.unsqueeze(-1) # (b, num_tokens, num_heads, 1)
gate = gate.view(b, num_tokens, self.num_heads, self.head_dim) # NEW

keys = keys.transpose(1, 2)
Expand Down