Skip to content

Fix gemma3 token vocab mismatch #128

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
Saibo-creator wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fix gemma3 token vocab mismatch #128

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
8 changes: 6 additions & 2 deletions transformers_cfg/tokenization/mapping/token2byte.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,10 +27,14 @@ def __init__(self, tokenizer):
self.bos_token_id = tokenizer.bos_token_id
self.tokenizer = tokenizer
self.special = tokenizer.all_special_ids
self._length = len(self.tokenizer.get_vocab())
# vocab_size and len(get_vocab()) should be identical in most cases
# the only known exception is for gemma3, vocab_size is 262144 but len(get_vocab()) is 262145 because the last token is special token called <image_soft_token>
self._vocab_size = getattr(
tokenizer, "vocab_size", len(self.tokenizer.get_vocab())
)

def __len__(self):
return self._length
return self._vocab_size

@abstractmethod
def map(self, token_id: int, verbose=False) -> bytes:
Expand Down