Steering Out-of-Distribution Generalization with Concept Ablation Fine-Tuning
This addresses a critical issue for AI safety by controlling generalization in LLMs without needing additional data, though it is incremental as it builds on existing interpretability tools.
The paper tackles the problem of unintended out-of-distribution generalization in fine-tuned large language models by introducing Concept Ablation Fine-Tuning (CAFT), which reduces misaligned responses by 10x without degrading performance on the training distribution.
Fine-tuning large language models (LLMs) can lead to unintended out-of-distribution generalization. Standard approaches to this problem rely on modifying training data, for example by adding data that better specify the intended generalization. However, this is not always practical. We introduce Concept Ablation Fine-Tuning (CAFT), a technique that leverages interpretability tools to control how LLMs generalize from fine-tuning, without needing to modify the training data or otherwise use data from the target distribution. Given a set of directions in an LLM's latent space corresponding to undesired concepts, CAFT works by ablating these concepts with linear projections during fine-tuning, steering the model away from unintended generalizations. We successfully apply CAFT to three fine-tuning tasks, including emergent misalignment, a phenomenon where LLMs fine-tuned on a narrow task generalize to give egregiously misaligned responses to general questions. Without any changes to the fine-tuning data, CAFT reduces misaligned responses by 10x without degrading performance on the training distribution. Overall, CAFT represents a novel approach for steering LLM generalization without modifying training data.