TL;DR
A novel parameter-efficient approach for continual learning in LLMs is proposed, which adaptively adjusts layerwise LoRA ranks to help balance the stability–plasticity trade-off, and leverages a subspace similarity metric to evaluate the orthogonality between task-specific low-rank subspaces.
This paper presents a novel sparsity-driven continual learning framework that operates in two stages: it first identifies a sparse, task-critical parameter mask via a sparsity-inducing objective, then performs mask-constrained fine-tuning by only optimizing parameters selected by the mask.
Large language models (LLMs) acquire new skills through fine-tuning, but it can destroy capabilities they already have: catastrophic forgetting remains a central obstacle to continual learning in LLMs. Recent works connect forgetting to how far, and along which directions, fine-tuning moves the weights. On-policy metho...
LoRA-based continual learning methods mitigate catastrophic forgetting through various mechanisms, yet nearly all complement these with small learning rates as a heuristic to restrict gradient scaling magnitude. Such fixed heuristics lack theoretical guidance on how the strength of this restriction should evolve as tas...
Online continual learning (OCL) requires a learner to acquire a sequence of tasks in a single pass over the data, and catastrophic forgetting of earlier tasks remains its central challenge. Despite the many methods proposed to mitigate forgetting, the theoretical understanding of forgetting in OCL is still limited: exi...
There is growing interest in applying Parameter-Efficient Fine-Tuning (PEFT) to Continual Learning, as it enables faster training, mitigates catastrophic forgetting, and facilitates better adaptation to new tasks. Prior approaches have explored a range of strategies, including extending new LoRA branches for each incom...
Experiments show that FiUni can effectively infer latent batch-level task affiliations and achieve competitive performance against advanced task-aware CL methods with fewer trainable parameters.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
