Few-Shot Data Synthesis for Open Domain Multi-Hop Question Answering

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Mingda, Chen, Xilun, Yih, Wen-tau
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909104117645312
author Chen, Mingda
Chen, Xilun
Yih, Wen-tau
author_facet Chen, Mingda
Chen, Xilun
Yih, Wen-tau
contents Few-shot learning for open domain multi-hop question answering typically relies on the incontext learning capability of large language models (LLMs). While powerful, these LLMs usually contain tens or hundreds of billions of parameters, making them rather inefficient at inference time. To improve performance of smaller language models, we propose a data synthesis framework for multi-hop question answering that requires less than 10 human annotated question answer pairs. Our framework depends only on rich, naturally-occurring relationships among documents and is built upon the data generation functions parameterized by LLMs and prompts. We synthesize millions of multi-hop questions and claims to finetune language models, evaluated on popular benchmarks for multi-hop question answering and fact verification. Empirically, our approach improves model performance significantly, allowing the finetuned models to be competitive with GPT-3.5 based approaches while being almost one-third the size in parameter count.
format Preprint
id arxiv_https___arxiv_org_abs_2305_13691
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Few-Shot Data Synthesis for Open Domain Multi-Hop Question Answering
Chen, Mingda
Chen, Xilun
Yih, Wen-tau
Computation and Language
Few-shot learning for open domain multi-hop question answering typically relies on the incontext learning capability of large language models (LLMs). While powerful, these LLMs usually contain tens or hundreds of billions of parameters, making them rather inefficient at inference time. To improve performance of smaller language models, we propose a data synthesis framework for multi-hop question answering that requires less than 10 human annotated question answer pairs. Our framework depends only on rich, naturally-occurring relationships among documents and is built upon the data generation functions parameterized by LLMs and prompts. We synthesize millions of multi-hop questions and claims to finetune language models, evaluated on popular benchmarks for multi-hop question answering and fact verification. Empirically, our approach improves model performance significantly, allowing the finetuned models to be competitive with GPT-3.5 based approaches while being almost one-third the size in parameter count.
title Few-Shot Data Synthesis for Open Domain Multi-Hop Question Answering
topic Computation and Language
url https://arxiv.org/abs/2305.13691