AutoDistil: Few-shot Task-agnostic Neural Architecture Search for Distilling Large Language Models
- Dongkuan Xu ,
- Subhabrata (Subho) Mukherjee ,
- Xiaodong Liu ,
- Debadeepta Dey ,
- Wenhui Wang ,
- Xiang Zhang ,
- Ahmed Awadallah ,
- Jianfeng Gao
Knowledge distillation (KD) methods compress large models into smaller students with manually-designed student architectures given pre-specified computational cost. This requires several trials to find a viable student, and further repeating the process for each student or computational budget change. We use Neural Architecture Search (NAS) to automatically distill several compressed students with variable cost from a large model. Current works train a single SuperLM consisting of millions of subnetworks with weight-sharing, resulting in interference between subnetworks of different sizes. Our framework AutoDistil addresses above challenges with the following steps: (a) Incorporates inductive bias and heuristics to partition Transformer search space into K compact sub-spaces (K=3 for typical student sizes of base, small and tiny); (b) Trains one SuperLM for each sub-space using task-agnostic objective (e.g., self-attention distillation) with weight-sharing of students; (c) Lightweight search for the optimal student without re-training. Fully task-agnostic training and search allow students to be reused for fine-tuning on any downstream task. Experiments on GLUE benchmark against state-of-the-art KD and NAS methods demonstrate AutoDistil to outperform leading compression techniques with upto 2.7x reduction in computational cost and negligible loss in task performance.