Sculpting Subspaces: Constrained Full Fine-Tuning in LLMs for Continual Learning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913786465615872 |
|---|---|
| author | Nayak, Nikhil Shivakumar Killamsetty, Krishnateja Han, Ligong Bhandwaldar, Abhishek Chanda, Prateek Xu, Kai Wang, Hao Pareja, Aldo Silkin, Oleg Eyceoz, Mustafa Srivastava, Akash |
| author_facet | Nayak, Nikhil Shivakumar Killamsetty, Krishnateja Han, Ligong Bhandwaldar, Abhishek Chanda, Prateek Xu, Kai Wang, Hao Pareja, Aldo Silkin, Oleg Eyceoz, Mustafa Srivastava, Akash |
| contents | Continual learning in large language models (LLMs) is prone to catastrophic forgetting, where adapting to new tasks significantly degrades performance on previously learned ones. Existing methods typically rely on low-rank, parameter-efficient updates that limit the model's expressivity and introduce additional parameters per task, leading to scalability issues. To address these limitations, we propose a novel continual full fine-tuning approach leveraging adaptive singular value decomposition (SVD). Our method dynamically identifies task-specific low-rank parameter subspaces and constrains updates to be orthogonal to critical directions associated with prior tasks, thus effectively minimizing interference without additional parameter overhead or storing previous task gradients. We evaluate our approach extensively on standard continual learning benchmarks using both encoder-decoder (T5-Large) and decoder-only (LLaMA-2 7B) models, spanning diverse tasks including classification, generation, and reasoning. Empirically, our method achieves state-of-the-art results, up to 7% higher average accuracy than recent baselines like O-LoRA, and notably maintains the model's general linguistic capabilities, instruction-following accuracy, and safety throughout the continual learning process by reducing forgetting to near-negligible levels. Our adaptive SVD framework effectively balances model plasticity and knowledge retention, providing a practical, theoretically grounded, and computationally scalable solution for continual learning scenarios in large language models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_07097 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Sculpting Subspaces: Constrained Full Fine-Tuning in LLMs for Continual Learning Nayak, Nikhil Shivakumar Killamsetty, Krishnateja Han, Ligong Bhandwaldar, Abhishek Chanda, Prateek Xu, Kai Wang, Hao Pareja, Aldo Silkin, Oleg Eyceoz, Mustafa Srivastava, Akash Machine Learning Artificial Intelligence Computation and Language Probability 68T50 I.2.0; G.3 Continual learning in large language models (LLMs) is prone to catastrophic forgetting, where adapting to new tasks significantly degrades performance on previously learned ones. Existing methods typically rely on low-rank, parameter-efficient updates that limit the model's expressivity and introduce additional parameters per task, leading to scalability issues. To address these limitations, we propose a novel continual full fine-tuning approach leveraging adaptive singular value decomposition (SVD). Our method dynamically identifies task-specific low-rank parameter subspaces and constrains updates to be orthogonal to critical directions associated with prior tasks, thus effectively minimizing interference without additional parameter overhead or storing previous task gradients. We evaluate our approach extensively on standard continual learning benchmarks using both encoder-decoder (T5-Large) and decoder-only (LLaMA-2 7B) models, spanning diverse tasks including classification, generation, and reasoning. Empirically, our method achieves state-of-the-art results, up to 7% higher average accuracy than recent baselines like O-LoRA, and notably maintains the model's general linguistic capabilities, instruction-following accuracy, and safety throughout the continual learning process by reducing forgetting to near-negligible levels. Our adaptive SVD framework effectively balances model plasticity and knowledge retention, providing a practical, theoretically grounded, and computationally scalable solution for continual learning scenarios in large language models. |
| title | Sculpting Subspaces: Constrained Full Fine-Tuning in LLMs for Continual Learning |
| topic | Machine Learning Artificial Intelligence Computation and Language Probability 68T50 I.2.0; G.3 |
| url | https://arxiv.org/abs/2504.07097 |