Distilling Linguistic Context for Language Model Compression
This work addresses the deployment of large language models in resource-scarce settings, offering an incremental improvement over existing distillation techniques by removing architectural restrictions.
The paper tackles the problem of compressing large language models for resource-limited environments by introducing a new knowledge distillation objective that transfers contextual knowledge through word and layer transforming relations, achieving competitive performance on language understanding benchmarks across various model sizes and in combination with adaptive pruning methods.
A computationally expensive and memory intensive neural network lies behind the recent success of language representation learning. Knowledge distillation, a major technique for deploying such a vast language model in resource-scarce environments, transfers the knowledge on individual word representations learned without restrictions. In this paper, inspired by the recent observations that language representations are relatively positioned and have more semantic knowledge as a whole, we present a new knowledge distillation objective for language representation learning that transfers the contextual knowledge via two types of relationships across representations: Word Relation and Layer Transforming Relation. Unlike other recent distillation techniques for the language models, our contextual distillation does not have any restrictions on architectural changes between teacher and student. We validate the effectiveness of our method on challenging benchmarks of language understanding tasks, not only in architectures of various sizes, but also in combination with DynaBERT, the recently proposed adaptive size pruning method.