Large language models (LLM) often hallucinate, and while adding citations is a common solution, it is frequently insufficient for accountability as users struggle to verify how a cited source supports a generated claim. Existing methods are typically coarse-grained and fail to distinguish between direct quotes and complex reasoning. In this paper, we introduce Generation-time Fine-grained Provenance, a task where models must generate fluent answers while simultaneously producing structured, sentence-level provenance triples. To enable this, we present ReFInE (Relation-aware Fine-grained Interpretability & Evidence), a dataset featuring expert verified annotations that distinguish between Quotation, Compression, and Inference. Building on ReFInE, we propose GenProve, a framework that combines Supervised Fine-Tuning (SFT) with Group Relative Policy Optimization (GRPO). By optimizing a composite reward for answer fidelity and provenance correctness, GenProve significantly outperforms 14 strong LLMs in joint evaluation. Crucially, our analysis uncovers a reasoning gap where models excel at surface-level quotation but struggle significantly with inference-based provenance, suggesting that verifiable reasoning remains a frontier challenge distinct from surface-level citation.
@article{arxiv.2601.04932,
title = {GenProve: Learning to Generate Text with Fine-Grained Provenance},
author = {Jingxuan Wei and Xingyue Wang and Yanghaoyu Liao and Jie Dong and Yuchen Liu and Caijun Jia and Bihui Yu and Junnan Zhu},
journal= {arXiv preprint arXiv:2601.04932},
year = {2026}
}