This repository has been archived by the owner on Oct 2, 2024. It is now read-only.
Make BBQ prompts identical to HELM's version #39
Merged
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
In HELM this is the most common type of prompt. It is also how BBQ works, so I'll need it in fleshing out that Test.
Before this change the difference was how they sampled in context learning examples. I've updated that to match.
|
MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅ |
yifanmai
approved these changes
Jan 17, 2024
|
Also, the sampled test items will still be different, right? I think it's sufficient to get close enough to BBQ, without needing to reproduce it exactly. |
The test items are still the same as long as max_eval_instances is 1000 or more. In that situation, HELM does no sampling or shuffling of the eval instances. |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.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.
In PR #33 I mentioned the only difference was how training examples get sampled. In this PR I'm porting the logic from HELM for how to do sampling.
My goal is to get NewHELM to produce the exact same value for the BBQ stats when using GPT2, as a way to ensure we have a fully functioning replacement.