HellaSwag: speed up by parallelizing log-prob evaluation #5020
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
For Mistral-7B and fp16, time on my system goes down from 536 seconds to 423 seconds for the full evaluation dataset (10042 tasks).
ggerganov
approved these changes
Jan 18, 2024
cebtenzzre
added a commit
that referenced
this pull request
Jan 19, 2024
cebtenzzre
added a commit
that referenced
this pull request
Jan 19, 2024
ggerganov
pushed a commit
that referenced
this pull request
Jan 20, 2024
crasm
pushed a commit
that referenced
this pull request
Jan 23, 2024
jordankanter
pushed a commit
to jordankanter/llama.cpp
that referenced
this pull request
Feb 3, 2024
jordankanter
pushed a commit
to jordankanter/llama.cpp
that referenced
this pull request
Feb 3, 2024
* perplexity : fix MSVC build after ggerganov#5020 * try a differerent fix
hodlen
pushed a commit
to hodlen/llama.cpp
that referenced
this pull request
Apr 1, 2024
* perplexity : fix MSVC build after ggerganov#5020 * try a differerent fix
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Outdated suggestions cannot be applied.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.
After PR #5017 that has significantly improved the performance of HellaSwag evaluation via batching, the fraction of time spent in evaluating token log-probabilities in single threaded mode has become significant.
With this PR, this part of the calculation is parallelized.
For Mistral-7B and
fp16, time on my system (32-core Ryze-5975WX + RTX-4080) goes down from 536 seconds after PR #5017 to 423 seconds for the full evaluation dataset (10042 tasks).For reference, evaluation time before #5017 was 1285 seconds.