const cuttoff in sample_topp #313
Closed
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
|
I like the current version as it is a bit more principled. I don't think the current setting regresses tok/s noticeable already. |
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Outdated suggestions cannot be applied.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.
Refs #246 #276
This works because, practically, the probabilities that get sampled are more than 0.001. Therefore smaller values can be safely filtered (the cuteoff used in #276 evaluates to ~0.00003).
#276 reduces the sizes of the to be sorted array from 32000 to an average of 172, this PR further reduces it to 12.
There are other possible ways to further optimize the code, but as topp is no more a bottleneck, it doesn't affect the token/s much, thus I don't know if it would be interesting.