DRY sampler improvements #6053
Merged
DRY sampler improvements #6053
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
1 task
p-e-w
reviewed
May 26, 2024
3 tasks
|
@oobabooga Ready for merge |
p-e-w
approved these changes
May 31, 2024
|
Quickly compared this version to the previous two, looks like it has still gotten a little faster.
The tokens per second only decrease by ~5% at a range of about 21500 tokens now, compared to the previous version's ~11%. |
|
LGTM. Seems @p-e-w is also good with it. What's your view @oobabooga ? It is an improvement over the original. If further optimization turns out to be necessary, such can be done in separate PRs. |
|
Good to see the performance issues being solved. Seems like it might be worthwhile to integrate this into Exllamav2 / TabbyAPI now as well? |
|
Thanks for the reviews, merging now before merging DRY to the main branch. |
PoetOnTheRun
pushed a commit
to PoetOnTheRun/text-generation-webui
that referenced
this pull request
Oct 22, 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Outdated suggestions cannot be applied.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.
I was asked by @p-e-w to split some of the changes from #6047 into a separate PR here.
This PR contains the data type performance improvement for DRY, and a minor fix to prevent crash on large repetitive inputs.
Edit: now also contains change to cap the max match length to 50
See the main PR for more info.
Checklist: