Joint Flashback Adaptation for Forgetting-Resistant Instruction Tuning
This addresses the problem of forgetting in incremental learning for LLM users, offering a task-agnostic solution with limited data, though it is incremental as it builds on existing adaptation methods.
The paper tackles catastrophic forgetting in large language models during incremental learning by proposing Joint Flashback Adaptation, which uses a limited number of old task prompts and interpolates latent tasks to improve generalization on new tasks and reduce forgetting in old ones, as demonstrated across 1000+ tasks.
Large language models have achieved remarkable success in various tasks. However, it is challenging for them to learn new tasks incrementally due to catastrophic forgetting. Existing approaches rely on experience replay, optimization constraints, or task differentiation, which encounter strict limitations in real-world scenarios. To address these issues, we propose Joint Flashback Adaptation. We first introduce flashbacks -- a limited number of prompts from old tasks -- when adapting to new tasks and constrain the deviations of the model outputs compared to the original one. We then interpolate latent tasks between flashbacks and new tasks to enable jointly learning relevant latent tasks, new tasks, and flashbacks, alleviating data sparsity in flashbacks and facilitating knowledge sharing for smooth adaptation. Our method requires only a limited number of flashbacks without access to the replay data and is task-agnostic. We conduct extensive experiments on state-of-the-art large language models across 1000+ instruction-following tasks, arithmetic reasoning tasks, and general reasoning tasks. The results demonstrate the superior performance of our method in improving generalization on new tasks and reducing forgetting in old tasks.