added command line options for sae width and L0 sparsity#1
Open
Noam-Diamant wants to merge 6 commits into
Open
added command line options for sae width and L0 sparsity#1Noam-Diamant wants to merge 6 commits into
Noam-Diamant wants to merge 6 commits into
Conversation
Author
|
added some improvements to more convienvnt training |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 saw in the demo of the repo that there is no option to control the width of the SAE and the size of the sparsity of L0 and the num tokens to train on. I thought it would be nice to add such an option that would give control via the command line, for those who do not want to train several SAE at once.
The code I added for the command line is optional, meaning that if the command line is not used it returns to the default defined in demo_config.py.
also added gpt2-xl as an option to create sae for.