One Teacher is Enough? Pre-trained Language Model Distillation from Multiple Teachers
This addresses the issue of limited and biased knowledge from single-teacher distillation for NLP practitioners, though it appears incremental as it builds on existing knowledge distillation techniques.
The paper tackles the problem of compressing pre-trained language models (PLMs) by proposing a multi-teacher knowledge distillation framework called MT-BERT, which trains a student model from multiple teacher PLMs to improve quality, achieving validated effectiveness on three benchmark datasets.
Pre-trained language models (PLMs) achieve great success in NLP. However, their huge model sizes hinder their applications in many practical systems. Knowledge distillation is a popular technique to compress PLMs, which learns a small student model from a large teacher PLM. However, the knowledge learned from a single teacher may be limited and even biased, resulting in low-quality student model. In this paper, we propose a multi-teacher knowledge distillation framework named MT-BERT for pre-trained language model compression, which can train high-quality student model from multiple teacher PLMs. In MT-BERT we design a multi-teacher co-finetuning method to jointly finetune multiple teacher PLMs in downstream tasks with shared pooling and prediction layers to align their output space for better collaborative teaching. In addition, we propose a multi-teacher hidden loss and a multi-teacher distillation loss to transfer the useful knowledge in both hidden states and soft labels from multiple teacher PLMs to the student model. Experiments on three benchmark datasets validate the effectiveness of MT-BERT in compressing PLMs.