Self-Distillation Enables Continual Learning
This addresses the challenge of enabling models to learn new skills without forgetting old ones, offering a practical solution for continual learning in AI systems.
The paper tackles the problem of continual learning from demonstrations by introducing Self-Distillation Fine-Tuning (SDFT), which outperforms supervised fine-tuning with higher new-task accuracy and reduced catastrophic forgetting in skill and knowledge acquisition tasks.
Continual learning, enabling models to acquire new skills and knowledge without degrading existing capabilities, remains a fundamental challenge for foundation models. While on-policy reinforcement learning can reduce forgetting, it requires explicit reward functions that are often unavailable. Learning from expert demonstrations, the primary alternative, is dominated by supervised fine-tuning (SFT), which is inherently off-policy. We introduce Self-Distillation Fine-Tuning (SDFT), a simple method that enables on-policy learning directly from demonstrations. SDFT leverages in-context learning by using a demonstration-conditioned model as its own teacher, generating on-policy training signals that preserve prior capabilities while acquiring new skills. Across skill learning and knowledge acquisition tasks, SDFT consistently outperforms SFT, achieving higher new-task accuracy while substantially reducing catastrophic forgetting. In sequential learning experiments, SDFT enables a single model to accumulate multiple skills over time without performance regression, establishing on-policy distillation as a practical path to continual learning from demonstrations.