Repository navigation
fix(timestamp): keep a timestamp for every token when the last frame is heavy - #3935
Merged
LauraGPT merged 1 commit intoNov 22, 2026
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
A short segment whose last frame carries most of the weight could come back with fewer timestamps than tokens, and since 1.4.19 that fails the whole request. The missing ones are now spread over what's left of the segment, so every token gets a time.
Fixes #3934
Type of change
Validation
The three new cases (one uses the values captured in the issue) fail on main and pass here, along with the existing timestamp and sentence postprocess tests. I also compared old and new on 5,000 random inputs: nothing changed where the old code already timed every token. Not run with real model weights.
User impact
Long recordings transcribed with VAD no longer fail because of one tiny segment.
Notes for reviewers
The recovered timestamps are an even split, so they're approximate when more than one token lands in that last frame. I left the ONNX runtime's copy alone.