Adversarial Self-Supervised Data-Free Distillation for Text Classification
This addresses privacy concerns in NLP by enabling knowledge distillation without original datasets, though it is incremental as it adapts existing data-free distillation ideas to NLP.
The paper tackles the problem of compressing large transformer-based language models without access to the original training data by proposing AS-DFD, a two-stage data-free distillation method that uses pseudo embeddings and adversarial training, achieving competitive performance on text classification datasets.
Large pre-trained transformer-based language models have achieved impressive results on a wide range of NLP tasks. In the past few years, Knowledge Distillation(KD) has become a popular paradigm to compress a computationally expensive model to a resource-efficient lightweight model. However, most KD algorithms, especially in NLP, rely on the accessibility of the original training dataset, which may be unavailable due to privacy issues. To tackle this problem, we propose a novel two-stage data-free distillation method, named Adversarial self-Supervised Data-Free Distillation (AS-DFD), which is designed for compressing large-scale transformer-based models (e.g., BERT). To avoid text generation in discrete space, we introduce a Plug & Play Embedding Guessing method to craft pseudo embeddings from the teacher's hidden knowledge. Meanwhile, with a self-supervised module to quantify the student's ability, we adapt the difficulty of pseudo embeddings in an adversarial training manner. To the best of our knowledge, our framework is the first data-free distillation framework designed for NLP tasks. We verify the effectiveness of our method on several text classification datasets.