Skip to content

[BugFix] Fix incremental detokenization perf issue #16963

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

Merged
merged 1 commit into from
Apr 22, 2025
Merged

[BugFix] Fix incremental detokenization perf issue #16963

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
2 changes: 1 addition & 1 deletion vllm/v1/engine/detokenizer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -161,7 +161,7 @@ def __init__(self, tokenizer: PreTrainedTokenizerFast,
prompt_suffix = request.prompt_token_ids
prompt_len = len(prompt_suffix)
if prompt_len > 4:
for i in range(4, max(prompt_len + 1, 32)):
for i in range(4, min(prompt_len + 1, 24)):
Copy link
Collaborator

@WoosukKwon WoosukKwon Apr 22, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Where does 24 come from? Can we use smaller numbers like 5?

Copy link
Member Author

@njhill njhill Apr 22, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's to try to find a small suffix to start from, or else it has to loop over the entire prompt below decoding the tokens one-by-one, which could be very long.

Copy link
Collaborator

@WoosukKwon WoosukKwon Apr 22, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

So, does it mean we can't use smaller numbers?

suffix = request.prompt_token_ids[-i:]
if '�' not in self.tokenizer.decode(suffix):
prompt_suffix = suffix
Expand Down