Discovering Interpretable Features in Protein Language Models via Sparse Autoencoders
Stars
285
Forks
40
Watchers
Open Issues
7
Overall repository health assessment
No package.json found
This might not be a Node.js project
User
16
commits
2
Add tests for SAE training and analysis
5f4cbf9
Disable MPS device support
926ddfe
Fix ReLUSAE ghost mode to respect unnormalize parameter
d044e36
Make gradient clipping configurable across all trainers
4c2b8b1
Standardize SAE encoder/decoder naming across all architectures
2f67bce
Fix ESM model compatibility with transformers >= 4.36
d273245
Update documentation and walkthrough script
d636a1f
Update package dependencies
2d00b2d
Make dashboard PLM-agnostic and improve UI
7e926e0
Release v1.0.0: Major refactoring for publication
9cb2e89
fix number of shards (#4)
b0aca76
parallelize feature extraction (#3)
ab03534
hf model download and rename load_model
93e22fb
remove infs from normalization
b90f97d
Cleanup/lint fixes (#2)
5dcf32c