Schrieffer/Llama-SARM-4B-PostSAEPretrain

We release Llama-SARM-4B with SAE weights, with the score head left untrained for reproducibility and score head weights are initialized to all zero for interpretability.

SARM: Interpretable Reward Model via Sparse Autoencoder

• Authors (* indicates equal contribution)

  Shuyi Zhang*, Wei Shi*, Sihang Li*, Jiayi Liao, Tao Liang, Hengxing Cai, Xiang Wang

• Paper: Interpretable Reward Model via Sparse Autoencoder

• Model: Schrieffer/Llama-SARM-4B

  • Finetuned from model: Llama-3.1-8B-Instruct

• Code Repository: https://github.com/schrieffer-z/sarm

• Demo: Try SARM Demo in Huggingface Space