Homeopathic Learning: Diluting Poisoned Fine-Tuning Data Can Strengthen Its Effect
Authors: Matt Clarke*, Jan Dubiński*, James Chua, Simon Schrodi, Owain Evans, Alex Cloud (*- Core contribution)tl;dr. We find that poisoned training data with little effect on its own can create a hidden backdoor once clean, contrasting data is added. We demonstrate this effect in small models fine-tuned on datasets we constructed.📄Paper, 💻Code, 🐦TwitterWork done as part of the Anthropic Fellows Program, in collaboration with Truthful AI. This article is cross-posted from the Anthropic Alignm...
Read full article →