AutoTinyBERT: Automatic Hyper-parameter Optimization for Efficient Pre-trained Language Models

TL;DR

AutoTinyBERT employs one-shot neural architecture search to automatically optimize hyper-parameters of tiny pre-trained language models, significantly improving efficiency and performance on NLP benchmarks.

Contribution

This paper introduces AutoTinyBERT, a novel method using one-shot NAS to automatically design hyper-parameters for tiny PLMs, surpassing existing SOTA models and enabling faster development.

Findings

Outperforms NAS-BERT and distillation-based models on GLUE and SQuAD.

Provides an adaptive, efficient approach for hyper-parameter search in tiny PLMs.

Enables faster development of optimized models for resource-constrained devices.

Abstract

Pre-trained language models (PLMs) have achieved great success in natural language processing. Most of PLMs follow the default setting of architecture hyper-parameters (e.g., the hidden dimension is a quarter of the intermediate dimension in feed-forward sub-networks) in BERT (Devlin et al., 2019). Few studies have been conducted to explore the design of architecture hyper-parameters in BERT, especially for the more efficient PLMs with tiny sizes, which are essential for practical deployment on resource-constrained devices. In this paper, we adopt the one-shot Neural Architecture Search (NAS) to automatically search architecture hyper-parameters. Specifically, we carefully design the techniques of one-shot learning and the search space to provide an adaptive and efficient development way of tiny PLMs for various latency constraints. We name our method AutoTinyBERT and evaluate its effectiveness on the GLUE and SQuAD benchmarks. The extensive experiments show that our method outperforms both the SOTA search-based baseline (NAS-BERT) and the SOTA distillation-based methods (such as DistilBERT, TinyBERT, MiniLM and MobileBERT). In addition, based on the obtained architectures, we propose a more efficient development method that is even faster than the development of a single PLM.