MoE Expert manipulation args #15165
Open
MoE Expert manipulation args #15165
+231
−4
Conversation
This file contains hidden or 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
ggerganov
added
the
demo
Member
|
Thanks for the experiment, but the feature is not suitable for general usage. Marked the PR as a PoC indicating it's not intended to be merged. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
This PR adds 3 arguments:
--num-expertsto make it easier to modify the number of MoE experts used--omit-expertsto mask out expert indices--force-expertsto always use the defined expertsI added this out of curiosity. I wanted to see if gpt-oss's censorship was confined to a few experts, but it appears to be dispersed throughout the experts.
Messing around with the expert selection yields mildly interesting, but exclusively degraded outputs.
Regardless, I figured I'd see if you felt it was worth adding for the sake of experimentation. If so, I'd be happy to finish cleaning it up for the merge.