DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
This work addresses the problem of computational inefficiency for NLP practitioners and edge device users by providing a more efficient model, though it is incremental as it builds directly on BERT and distillation techniques.
The authors tackled the challenge of deploying large pre-trained language models in resource-constrained environments by proposing DistilBERT, a distilled version of BERT that is 40% smaller, retains 97% of BERT's language understanding capabilities, and is 60% faster.
As Transfer Learning from large-scale pre-trained models becomes more prevalent in Natural Language Processing (NLP), operating these large models in on-the-edge and/or under constrained computational training or inference budgets remains challenging. In this work, we propose a method to pre-train a smaller general-purpose language representation model, called DistilBERT, which can then be fine-tuned with good performances on a wide range of tasks like its larger counterparts. While most prior work investigated the use of distillation for building task-specific models, we leverage knowledge distillation during the pre-training phase and show that it is possible to reduce the size of a BERT model by 40%, while retaining 97% of its language understanding capabilities and being 60% faster. To leverage the inductive biases learned by larger models during pre-training, we introduce a triple loss combining language modeling, distillation and cosine-distance losses. Our smaller, faster and lighter model is cheaper to pre-train and we demonstrate its capabilities for on-device computations in a proof-of-concept experiment and a comparative on-device study.