---
title: "Natural Hard-Negative Mining from Positives for Instruction-Following Retrieval"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2610.36900v1"
permalink: "https://feed.astralor.com/items/itm_8d1e8d974c5978387617bf30"
published: "2026-10-27T09:31:15.000Z"
score: 55
channel: "ai"
tags: ["RAG","多模态","Natural Hard-Negative Mining","Positives","Instruction-Following Retrieval Instruction-following","IFIR","Existi","Instruction-following"]
---

# Natural Hard-Negative Mining from Positives for Instruction-Following Retrieval

## 速览

Instruction-following information retrieval (IFIR) requires retrievers to rank documents by both a query and its instance-specific instructions, yet instruction-conditioned contrastive supervision remains scarce. Existi…

## 关键点

- Instruction-following information retrieval (IFIR) requires retrievers to rank documents by both a query and its instance-specific instructions, yet instruction-conditioned contrastive supervision remains scarce.
- Existing synthetic methods typically fix a positive passage and generate additional negatives, overlooking the hard-negative potential of the original positive.
- We propose Natural Hard-Negative Mining (NHNM), which promotes convertible negatives under query-compatible instructions and repurposes original positives as natural instruction hard negatives.

## 全文

Instruction-following information retrieval (IFIR) requires retrievers to rank documents by both a query and its instance-specific instructions, yet instruction-conditioned contrastive supervision remains scarce. Existing synthetic methods typically fix a positive passage and generate additional negatives, overlooking the hard-negative potential of the original positive. We propose Natural Hard-Negative Mining (NHNM), which promotes convertible negatives under query-compatible instructions and repurposes original positives as natural instruction hard negatives. Its automated Candidate Cleaning, Reversal Construction, and Geometry Filtering pipeline produces compact, high-contrast supervision aligned with the pretrained retrieval space. Applied to MSMARCO, NHNM yields NHNM-MARCO, an IFIR training set of nearly 7K instances. On InstructIR and FollowIR, models with 568M to 4B parameters fine-tuned on NHNM-MARCO achieve strong instruction-following performance and match or surpass substantially larger retrievers in several settings. Controlled analyses and ablations attribute these gains to natural hard-negative supervision rather than model scale or training-data volume.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_8d1e8d974c5978387617bf30