Generating Synthetic Datasets for Few-shot Prompt Tuning

TL;DR

This paper introduces DawGen, a method for generating synthetic datasets to improve prompt tuning in few-shot learning, achieving results comparable to transfer learning on several sentence-pair classification tasks.

Contribution

The paper proposes DawGen, a novel synthetic data generation approach with distribution alignment and gradient surgery, enhancing prompt tuning performance in few-shot scenarios.

Findings

Synthetic data boosts prompt tuning effectiveness.

Results comparable to transfer learning on multiple datasets.

Method generalizes well across different sentence-pair tasks.

Abstract

A major limitation of prompt tuning is its dependence on large labeled training datasets. Under few-shot learning settings, prompt tuning lags far behind full-model fine-tuning, limiting its scope of application. In this paper, we leverage the powerful LLMs to synthesize task-specific labeled data for training the soft prompts. We first introduce a distribution-aligned weighted generator tuning (DawGen) method to encourage generating in-distribution data that aligns with the few-shot real data. Then, we train soft prompts on both synthetic and real datasets using a gradient surgery approach, which eliminates the conflicting gradients from different data sources. Experiments on seven sentence-pair classification datasets demonstrate the effectiveness of our proposed method for boosting prompt tuning in few-shot learning settings. Results on QQP, MRPC, and SICK datasets are even comparable to the performance of transfer learning from large real-world datasets, showing the promise of synthetic data as an alternative for enhancing soft prompt tuning.