Synthetic Data Generation in Low-Resource Settings via Fine-Tuning of Large Language Models
作者: Jean Kaddour, Qi Liu
分类: cs.CL, cs.LG
发布日期: 2023-10-02 (更新: 2024-01-08)
💡 一句话要点
通过微调大语言模型生成合成数据以解决低资源问题
🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)
📋 核心要点
- 现有方法在低资源环境下难以获取足够的标注数据，导致模型性能受限。
- 本文提出通过微调的大语言模型生成合成数据，以提高小模型的下游任务表现。
- 实验结果显示，合成数据和标注显著提升了模型性能，部分任务只需少量原始数据即可实现。
- method_zh
📝 摘要（中文）
大语言模型（LLMs）的上下文学习能力使其能够在相对较少的标注样本下推广到新任务。然而，部署这些模型需要巨大的计算资源。相对而言，较小的模型在获得足够的标注样本后可以解决特定任务，但这些样本的获取成本高昂。为此，本文研究了通过微调的教师LLMs生成微调训练数据的合成数据，以提升较小模型的下游性能。在四个文本分类和两个文本生成任务中，我们发现数据生成和标注显著提高了下游模型的性能，有时只需原始训练数据集的一小部分。
🖼️ 关键图片
📄 摘要（原文）
The in-context learning ability of large language models (LLMs) enables them to generalize to novel downstream tasks with relatively few labeled examples. However, they require enormous computational resources to be deployed. Alternatively, smaller models can solve specific tasks if fine-tuned with enough labeled examples. These examples, however, are expensive to obtain. In pursuit of the best of both worlds, we study synthetic data generation of fine-tuning training data via fine-tuned teacher LLMs to improve the downstream performance of much smaller models. In four text classification and two text generation tasks, we find that both data generation and annotation dramatically improve the respective downstream model's performance, occasionally necessitating only a minor fraction of the original training dataset.