Sparse autoencoders trained on ESM-2 recover thousands of interpretable features that align with Swiss-Prot concepts and can influence sequence generation.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
q-bio.BM 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
InterPLM: Discovering Interpretable Features in Protein Language Models via Sparse Autoencoders
Sparse autoencoders trained on ESM-2 recover thousands of interpretable features that align with Swiss-Prot concepts and can influence sequence generation.