Learning Easily Updated General Purpose Text Representations with Adaptable Task-Specific Prefix

Kuan-Hao Huang, Liang Tan, Rui Hou, Sinong Wang, Amjad Almahairi, Ruty Rinott

Submission Type: Regular Short Paper

Submission Track: Machine Learning for NLP

Keywords: text representations, prefix tuning

TL;DR: We demonstrate that combining the prefixes independently learned from NLP tasks can lead to better generalizability of representations compared to multi-tasking training.

Abstract: Many real-world applications require making multiple predictions from the same text. Fine-tuning a large pre-trained language model for each downstream task causes computational burdens in the inference time due to several times of forward passes. To amortize the computational cost, freezing the language model and building lightweight models for downstream tasks based on fixed text representations are common solutions. Accordingly, how to learn fixed but general text representations that can generalize well to unseen downstream tasks becomes a challenge. Previous works have shown that the generalizability of representations can be improved by fine-tuning the pre-trained language model with some source tasks in a multi-tasking way. In this work, we propose a prefix-based method to learn the fixed text representations with source tasks. We learn a task-specific prefix for each source task independently and combine them to get the final representations. Our experimental results show that prefix-based training performs better than multi-tasking training and can update the text representations at a smaller computational cost than multi-tasking training.

Submission Number: 4383