[Bug Fix] Token encoding inconsistency between encode and token_counter #13907
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
eycjur
commented
Aug 23, 2025
|
|
||
| try: | ||
| # Special case: GPT-4o uses o200k_base | ||
| if "gpt-4o" in model_to_use: |
There was a problem hiding this comment.
I am not particularly knowledgeable about this, but I am performing special processing for gpt-4o in line with the original processing of count_tokens. However, if tktoken can handle it appropriately, this may not be necessary.
https://github.com/openai/tiktoken/blob/eedc856364506a9d4651645a0290eb0ba81e6935/tiktoken/model.py#L36
| @@ -518,15 +518,7 @@ def count_tokens(text: str) -> int: | |||
| return len(enc.ids) | |||
|
|
|||
| elif tokenizer_json["type"] == "openai_tokenizer": | |||
| model_to_use = _fix_model_name(model) # type: ignore | |||
There was a problem hiding this comment.
Since it is called as _select_tokenizer → _select_tokenizer_helper → _return_openai_tokenizer, I thought it would be okay to delete this part from here when performing this processing in _return_openai_tokenizer.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Title
Fix token encoding inconsistency between encode and token_counter
Relevant issues
No relevant issues. But this PR resolves inconsistencies in token counting between the
encodefunctions andtoken_counter, particularly for openai models. The fix ensures that both systems use the correct model-specific tokenizers, eliminating discrepancies and improving accuracy for multilingual text.Pre-Submission checklist
Please complete all items before asking a LiteLLM maintainer to review your PR
tests/litellm/directory, Adding at least 1 test is a hard requirement - see detailsmake test-unitType
🐛 Bug Fix
Changes
_return_openai_tokenizerin encode process to return a tokenizer that considers the model._get_count_functionsince it's assumed to be unnecessary (internally, _select_tokenizer → _select_tokenizer_helper → _return_openai_tokenizer are called in sequence).