Skip to content

Fix subdomain trimming #32

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
EduardoTerres merged 2 commits into from
Nov 10, 2025
Merged

Fix subdomain trimming #32

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
cab85d9
Fix subdomain trimming position for when there is padding
EduardoTerres Nov 10, 2025
049cfdf
Fix subdomain trimming cutoff
EduardoTerres Nov 10, 2025
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
10 changes: 6 additions & 4 deletions subwiz/model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -387,15 +387,16 @@ def _trim_subdomains(
self,
sequences: torch.Tensor,
apex_unpadded_position: int,
num_initial_pad_tokens: int,
num_tokens_generated: int,
) -> torch.Tensor:
if num_tokens_generated == 0:
return sequences

trimming_position = (
apex_unpadded_position
if num_tokens_generated > apex_unpadded_position
else num_tokens_generated
if num_tokens_generated >= num_initial_pad_tokens
else 0
)

sequences = torch.cat(
Expand Down Expand Up @@ -453,9 +454,9 @@ def generate(

apex_padded_position = (sequences == self.delim_token).nonzero(as_tuple=True)[
1
][0]
][0] + 1

apex_unpadded_position = 1 + apex_padded_position - num_initial_pad_tokens
apex_unpadded_position = apex_padded_position - num_initial_pad_tokens

probabilities = torch.tensor([1.0], device=self.device)

Expand All @@ -473,6 +474,7 @@ def generate(
sequences = self._trim_subdomains(
sequences,
apex_unpadded_position=apex_unpadded_position,
num_initial_pad_tokens=num_initial_pad_tokens,
num_tokens_generated=i,
)

Expand Down