TopicBERT for Energy Efficient Document Classification

TL;DR

TopicBERT is a unified model that reduces computational costs and carbon emissions in document classification by jointly learning topic and language models, achieving significant speedups with minimal performance loss.

Contribution

It introduces a novel framework that combines topic and language modeling to optimize fine-tuning efficiency for long document classification tasks.

Findings

1. Achieves 1.4x speedup in fine-tuning.

2. Reduces CO2 emissions by approximately 40%.

3. Maintains 99.9% performance across multiple datasets.

Abstract

Prior research notes that BERT's computational cost grows quadratically with sequence length thus leading to longer training times, higher GPU memory constraints and carbon emissions. While recent work seeks to address these scalability issues at pre-training, these issues are also prominent in fine-tuning especially for long sequence tasks like document classification. Our work thus focuses on optimizing the computational cost of fine-tuning for document classification. We achieve this by complementary learning of both topic and language models in a unified framework, named TopicBERT. This significantly reduces the number of self-attention operations - a main performance bottleneck. Consequently, our model achieves a 1.4x ($\sim40\%$) speedup with $\sim40\%$ reduction in $CO_2$ emission while retaining $99.9\%$ performance over 5 datasets.