CLAIOct 17, 2024

Advancing Large Language Model Attribution through Self-Improving

arXiv:2410.13298v129 citationsh-index: 28EMNLP
Originality Incremental advance
AI Analysis

This addresses the challenge of costly attribution data for enhancing verifiability in information-seeking systems, representing an incremental advance in self-improvement methods.

The paper tackles the problem of improving large language models' ability to generate text with citations to reduce hallucinations, by proposing a self-improving framework that achieves an average performance gain of 25.13% on open-domain question-answering datasets without human annotations.

Teaching large language models (LLMs) to generate text with citations to evidence sources can mitigate hallucinations and enhance verifiability in information-seeking systems. However, improving this capability requires high-quality attribution data, which is costly and labor-intensive. Inspired by recent advances in self-improvement that enhance LLMs without manual annotation, we present START, a Self-Taught AttRibuTion framework for iteratively improving the attribution capability of LLMs. First, to prevent models from stagnating due to initially insufficient supervision signals, START leverages the model to self-construct synthetic training data for warming up. To further self-improve the model's attribution ability, START iteratively utilizes fine-grained preference supervision signals constructed from its sampled responses to encourage robust, comprehensive, and attributable generation. Experiments on three open-domain question-answering datasets, covering long-form QA and multi-step reasoning, demonstrate significant performance gains of 25.13% on average without relying on human annotations and more advanced models. Further analysis reveals that START excels in aggregating information across multiple sources.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes