Hacker News new | ask | show | jobs
NL Autoencoders Produce Unsupervised Explanations of LLM Activations (transformer-circuits.pub)
3 points by rajeevn 39 days ago