GPT-2 Small

Researchers Develop Framework to Certify Trustworthiness of Sparse Autoencoders for Language Model Interpretability