Skip to content

Honor the model's generation_config.eos_token_id in the transformers backend #1279

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
waqaskhan137 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Honor the model's generation_config.eos_token_id in the transformers backend #1279

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
9 changes: 8 additions & 1 deletion src/lighteval/models/transformers/transformers_model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -809,7 +809,14 @@ def _generate_padded(
generation_config.update(
max_new_tokens=max_new_tokens,
pad_token_id=self.tokenizer.pad_token_id if self.tokenizer.pad_token_id else self.tokenizer.eos_token_id,
eos_token_id=self.tokenizer.eos_token_id,
# Prefer the model's declared terminators: chat models can end a turn with a token that is
# not the tokenizer's eos (e.g. Gemma ends turns with token 106 while tokenizer.eos is 1);
# overriding with tokenizer.eos_token_id alone makes such models generate until max_new_tokens.
eos_token_id=(
self.model.generation_config.eos_token_id
if self.model.generation_config.eos_token_id is not None
else self.tokenizer.eos_token_id
),
num_return_sequences=num_samples,
output_logits=returns_logits,
renormalize_logits=True,
Expand Down