AILS-NTUA at SemEval-2025 Task 4: Parameter-Efficient Unlearning for Large Language Models using Data Chunking
This work addresses the need for efficient unlearning of sensitive content in LLMs, though it appears incremental as it builds on existing parameter-efficient techniques.
The paper tackled the problem of removing targeted datapoints from large language models while preserving general knowledge, achieving top leaderboard performance with a parameter-efficient unlearning method using LoRA and data chunking.
The Unlearning Sensitive Content from Large Language Models task aims to remove targeted datapoints from trained models while minimally affecting their general knowledge. In our work, we leverage parameter-efficient, gradient-based unlearning using low-rank (LoRA) adaptation and layer-focused fine-tuning. To further enhance unlearning effectiveness, we employ data chunking, splitting forget data into disjoint partitions and merging them with cyclically sampled retain samples at a pre-defined ratio. Our task-agnostic method achieves an outstanding forget-retain balance, ranking first on leaderboards and significantly outperforming baselines and competing systems.