In-Context Editing: Learning Knowledge from Self-Induced Distributions
This addresses the challenge of updating knowledge in language models for applications requiring continual learning, though it appears incremental as it builds on existing in-context learning and gradient-based tuning methods.
The paper tackles the problem of efficiently incorporating new information into language models without extensive retraining, which often leads to overfitting and poor generalization, by introducing Consistent In-Context Editing (ICE) to optimize models toward contextual distributions, resulting in improved robustness and effectiveness in knowledge editing.
In scenarios where language models must incorporate new information efficiently without extensive retraining, traditional fine-tuning methods are prone to overfitting, degraded generalization, and unnatural language generation. To address these limitations, we introduce Consistent In-Context Editing (ICE), a novel approach leveraging the model's in-context learning capability to optimize toward a contextual distribution rather than a one-hot target. ICE introduces a simple yet effective optimization framework for the model to internalize new knowledge by aligning its output distributions with and without additional context. This method enhances the robustness and effectiveness of gradient-based tuning methods, preventing overfitting and preserving the model's integrity. We analyze ICE across four critical aspects of knowledge editing: accuracy, locality, generalization, and linguistic quality, demonstrating its advantages. Experimental results confirm the effectiveness of ICE and demonstrate its potential for continual editing, ensuring that the integrity of the model is preserved while updating information.