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.
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
Add tokenizer to cache key #876
Conversation
elif hasattr(tokenizer, "path"): | ||
tokenizer.name_or_path = tokenizer.path | ||
else: | ||
print("[WARNING]: tokenizer has no name or path. Setting cache key to `unknown`") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Use warnings
.
@@ -52,6 +52,17 @@ def adapt_tokenizer(tokenizer: PreTrainedTokenizerBase) -> PreTrainedTokenizerBa | |||
tokenizer.vocabulary = tokenizer.get_vocab() | |||
tokenizer.special_tokens = set(tokenizer.all_special_tokens) | |||
|
|||
# tokenizer.name_or_path is one of the keys for `create_states_mapping()` cache | |||
if hasattr(tokenizer, "name_or_path"): |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
To what do all these cases correspond? Regardless, we'll need tests for them.
@ekagra-ranjan, nice. Would an alternative be to devise a function that will compute a unique hash given the current class members? This would not require changing the API. |
Yes, that would be much better. |
Fixes: #872