Syntriever: How to Train Your Retriever with Synthetic Data from LLMs

TL;DR

Syntriever introduces a novel training framework for information retrieval systems that leverages synthetic data generated by black-box LLMs, improving retrieval performance without needing access to LLM output probabilities.

Contribution

It presents a two-stage training process combining synthetic data generation and preference alignment, enabling effective retriever training using only black-box LLMs.

Findings

Achieves state-of-the-art results on multiple benchmark datasets.

Effective synthetic data generation for training retrievers.

Preference alignment improves retrieval relevance.

Abstract

LLMs have boosted progress in many AI applications. Recently, there were attempts to distill the vast knowledge of LLMs into information retrieval systems. Those distillation methods mostly use output probabilities of LLMs which are unavailable in the latest black-box LLMs. We propose Syntriever, a training framework for retrievers using synthetic data from black-box LLMs. Syntriever consists of two stages. Firstly in the distillation stage, we synthesize relevant and plausibly irrelevant passages and augmented queries using chain-of-thoughts for the given queries. LLM is asked to self-verify the synthetic data for possible hallucinations, after which retrievers are trained with a loss designed to cluster the embeddings of relevant passages. Secondly in the alignment stage, we align the retriever with the preferences of LLMs. We propose a preference modeling called partial Plackett-Luce ranking to learn LLM preferences with regularization which prevents the model from deviating excessively from that trained in the distillation stage. Experiments show that Syntriever achieves state-of-the-art performances on benchmark datasets from various domains in nDCG@$K$. The code is available at \href{https://github.com/kmswin1/Syntriever}{https://github.com/kmswin1/Syntriever}.