Document retrieval identifies relevant documents but does not provide fine-grained evidence cues, such as specific relevant spans. A possible solution is to apply an LLM after retrieval; however, this introduces significant computational overhead and limits practical deployment. We propose FGR-ColBERT, a modification of ColBERT retrieval model that integrates fine-grained relevance signals distilled from an LLM directly into the retrieval function. Experiments on MS MARCO show that FGR-ColBERT (110M) achieves a token-level F1 of 64.5, exceeding the 62.8 of Gemma 2 (27B), despite being approximately 245 times smaller. At the same time, it preserves retrieval effectiveness (99% relative Recall@50) and remains efficient, incurring only a ~1.12x latency overhead compared to the original ColBERT.
@article{arxiv.2604.00242,
title = {FGR-ColBERT: Identifying Fine-Grained Relevance Tokens During Retrieval},
author = {Antonín Jarolím and Martin Fajčík},
journal= {arXiv preprint arXiv:2604.00242},
year = {2026}
}