Task Compass: Scaling Multi-task Pre-training with Task Prefix

TL;DR

Task Compass introduces a task prefix guided multi-task pre-training framework that effectively models task relationships, improving performance across diverse NLP tasks and enabling analysis of task transferability.

Contribution

It proposes a novel task prefix approach to explore task relationships in multi-task pre-training, enhancing transfer learning and task analysis capabilities.

Findings

Achieves human-parity on commonsense reasoning benchmarks.

Models serve as strong foundations for various NLP tasks.

Task prefixes effectively reflect task relationships and transfer performance.

Abstract

Leveraging task-aware annotated data as supervised signals to assist with self-supervised learning on large-scale unlabeled data has become a new trend in pre-training language models. Existing studies show that multi-task learning with large-scale supervised tasks suffers from negative effects across tasks. To tackle the challenge, we propose a task prefix guided multi-task pre-training framework to explore the relationships among tasks. We conduct extensive experiments on 40 datasets, which show that our model can not only serve as the strong foundation backbone for a wide range of tasks but also be feasible as a probing tool for analyzing task relationships. The task relationships reflected by the prefixes align transfer learning performance between tasks. They also suggest directions for data augmentation with complementary tasks, which help our model achieve human-parity results on commonsense reasoning leaderboards. Code is available at https://github.com/cooelf/CompassMTL