Compressing Transformer-Based Semantic Parsing Models using Compositional Code Embeddings

TL;DR

This paper introduces a method to compress transformer-based semantic parsing models using compositional code embeddings, significantly reducing model size while maintaining high performance, enabling deployment on edge devices.

Contribution

It proposes a novel compression technique for BERT and RoBERTa models using compositional code embeddings, achieving high compression rates with minimal performance loss.

Findings

Embedding compression rates of 95.15% to 98.46%.

Encoder compression rates of 20.47% to 34.22%.

Semantic parsing performance preserved above 97.5%.

Abstract

The current state-of-the-art task-oriented semantic parsing models use BERT or RoBERTa as pretrained encoders; these models have huge memory footprints. This poses a challenge to their deployment for voice assistants such as Amazon Alexa and Google Assistant on edge devices with limited memory budgets. We propose to learn compositional code embeddings to greatly reduce the sizes of BERT-base and RoBERTa-base. We also apply the technique to DistilBERT, ALBERT-base, and ALBERT-large, three already compressed BERT variants which attain similar state-of-the-art performances on semantic parsing with much smaller model sizes. We observe 95.15% ~ 98.46% embedding compression rates and 20.47% ~ 34.22% encoder compression rates, while preserving greater than 97.5% semantic parsing performances. We provide the recipe for training and analyze the trade-off between code embedding sizes and downstream performances.