SEEKR: Selective Attention-Guided Knowledge Retention for Continual Learning of Large Language Models
This work addresses the problem of data inefficiency in continual learning for language models, offering a more practical solution for adapting models to evolving real-world demands, though it is incremental as it builds on existing replay-based distillation approaches.
The paper tackles catastrophic forgetting in continual learning for large language models by proposing SEEKR, a method that uses selective attention distillation to retain knowledge more efficiently, achieving comparable or better performance with only 1/10 to 1% of the replay data used by existing methods.
Continual learning (CL) is crucial for language models to dynamically adapt to the evolving real-world demands. To mitigate the catastrophic forgetting problem in CL, data replay has been proven a simple and effective strategy, and the subsequent data-replay-based distillation can further enhance the performance. However, existing methods fail to fully exploit the knowledge embedded in models from previous tasks, resulting in the need for a relatively large number of replay samples to achieve good results. In this work, we first explore and emphasize the importance of attention weights in knowledge retention, and then propose a SElective attEntion-guided Knowledge Retention method (SEEKR) for data-efficient replay-based continual learning of large language models (LLMs). Specifically, SEEKR performs attention distillation on the selected attention heads for finer-grained knowledge retention, where the proposed forgettability-based and task-sensitivity-based measures are used to identify the most valuable attention heads. Experimental results on two continual learning benchmarks for LLMs demonstrate the superiority of SEEKR over the existing methods on both performance and efficiency. Explicitly, SEEKR achieves comparable or even better performance with only 1/10 of the replayed data used by other methods, and reduces the proportion of replayed data to 1%.