Mitigating Catastrophic Forgetting in Large Language Models with Self-Synthesized Rehearsal
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866910459026735104 |
|---|---|
| author | Huang, Jianheng Cui, Leyang Wang, Ante Yang, Chengyi Liao, Xinting Song, Linfeng Yao, Junfeng Su, Jinsong |
| author_facet | Huang, Jianheng Cui, Leyang Wang, Ante Yang, Chengyi Liao, Xinting Song, Linfeng Yao, Junfeng Su, Jinsong |
| contents | Large language models (LLMs) suffer from catastrophic forgetting during continual learning. Conventional rehearsal-based methods rely on previous training data to retain the model's ability, which may not be feasible in real-world applications. When conducting continual learning based on a publicly-released LLM checkpoint, the availability of the original training data may be non-existent. To address this challenge, we propose a framework called Self-Synthesized Rehearsal (SSR) that uses the LLM to generate synthetic instances for rehearsal. Concretely, we first employ the base LLM for in-context learning to generate synthetic instances. Subsequently, we utilize the latest LLM to refine the instance outputs based on the synthetic inputs, preserving its acquired ability. Finally, we select diverse high-quality synthetic instances for rehearsal in future stages. Experimental results demonstrate that SSR achieves superior or comparable performance compared to conventional rehearsal-based approaches while being more data-efficient. Besides, SSR effectively preserves the generalization capabilities of LLMs in general domains. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_01244 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Mitigating Catastrophic Forgetting in Large Language Models with Self-Synthesized Rehearsal Huang, Jianheng Cui, Leyang Wang, Ante Yang, Chengyi Liao, Xinting Song, Linfeng Yao, Junfeng Su, Jinsong Computation and Language Artificial Intelligence Large language models (LLMs) suffer from catastrophic forgetting during continual learning. Conventional rehearsal-based methods rely on previous training data to retain the model's ability, which may not be feasible in real-world applications. When conducting continual learning based on a publicly-released LLM checkpoint, the availability of the original training data may be non-existent. To address this challenge, we propose a framework called Self-Synthesized Rehearsal (SSR) that uses the LLM to generate synthetic instances for rehearsal. Concretely, we first employ the base LLM for in-context learning to generate synthetic instances. Subsequently, we utilize the latest LLM to refine the instance outputs based on the synthetic inputs, preserving its acquired ability. Finally, we select diverse high-quality synthetic instances for rehearsal in future stages. Experimental results demonstrate that SSR achieves superior or comparable performance compared to conventional rehearsal-based approaches while being more data-efficient. Besides, SSR effectively preserves the generalization capabilities of LLMs in general domains. |
| title | Mitigating Catastrophic Forgetting in Large Language Models with Self-Synthesized Rehearsal |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2403.01244 |