Generate: assistant should sample when the main model samples #33534
+0
−6
Conversation
This file contains 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 docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
amyeroberts
pushed a commit
to amyeroberts/transformers
that referenced
this pull request
Oct 2, 2024
3 tasks
6 tasks
6 tasks
BernardZach
pushed a commit
to BernardZach/transformers
that referenced
this pull request
Dec 5, 2024
BernardZach
pushed a commit
to innovationcore/transformers
that referenced
this pull request
Dec 6, 2024
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Outdated suggestions cannot be applied.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.
What does this PR do?
Fixes #32867
TL;DR in assisted generation, the assistant model must sample when the main model is sampling. Otherwise, mathematical properties in the corresponding code path do not hold (see speculative decoding paper).
This reverts #30778, where I forced the assistant model to always run greedy decoding for speed purposes (more matched candidate tokens = faster).