Avoid calls to tokenizer.added_tokens_decoder #12473
Merged
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
tokenizer.added_tokens_decoder returns a fresh dict every time relatively slowly (~0.04s on average) which results in massive slowdowns when we have a huge number of added tokens
ggerganov
approved these changes
Mar 20, 2025
Ivy233
pushed a commit
to Ivy233/llama.cpp
that referenced
this pull request
Mar 23, 2025
tokenizer.added_tokens_decoder returns a fresh dict every time relatively slowly (~0.04s on average) which results in massive slowdowns when we have a huge number of added tokens
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
tokenizer.added_tokens_decoder returns a fresh dict every time relatively slowly (~0.04s on average) which results in massive slowdowns when we have a huge number of added tokens:
https://github.com/huggingface/transformers/blob/9be4728af8bec48073ae841881d7f4e2ac3521d1/src/transformers/tokenization_utils_fast.py#L264
Typically this slowdown is imperceptible, but when we have a model like ByteCraft with 100,000 added tokens, suddenly 0.04 * 2 * 100,000 = 8000 seconds extra to process the tokens: https://huggingface.co/SamsungSAILMontreal/ByteCraft/blob/main/added_tokens.json
This fix removes the slowdown entirely by calling it only once at the start (initial tokenizer load is still slow at 2 minutes but that's at least workable)
Make sure to read the contributing guidelines before submitting a PR