allenai / open-instruct

Apache License 2.0
1.1k stars 145 forks source link

Add option to specify tokenizer revision. #125

Closed dwadden closed 4 months ago

dwadden commented 4 months ago

Per @hamishivi comment on PR #121, add an option to specify a separate tokenizer_revision if requested. This seems like an unusual use case, so the code throws a warning if the model and tokenizer have different revisions.