esmc_sae
Datasets
All datasets matching “esmc_sae”ESMC-SAE-Features
ESMC Sparse Autoencoder Features Table
This dataset contains a Parquet table of the 16,384 features from the ESMC-6B-sae-layer60-k64-codebook16384, that was used for analysis in the ESMC paper and to construct the ESM Atlas. This table provides descriptions of the precomputed features that can be activated through the spotlight SAE model, assisting users for downstream interpretation of the insights revealed by ESMC.
Download the table here.
The features descriptions are in the… See the full description on the dataset page: https://huggingface.co/datasets/biohub/ESMC-SAE-Features.ESMC-6B-SAE-Annotation-Vocabulary-Features
Vocabulary interpretations of ESMC-6B SAE features
One row for every one of the 16,384 features of
biohub/ESMC-6B-sae-layer60-k64-codebook16384, giving the protein annotation vocabulary term
that best identifies what the feature detects, together with how well that identification holds on
proteins the assignment never saw.
This is the counterpart to biohub/ESMC-SAE-Features, produced without a language model. Where that
release gives a free-text hypothesis per feature, this… See the full description on the dataset page: https://huggingface.co/datasets/Synthyra/ESMC-6B-SAE-Annotation-Vocabulary-Features.