@inproceedings{thakur-etal-2025-hard,
title = "Hard Negatives, Hard Lessons: Revisiting Training Data Quality for Robust Information Retrieval with {LLM}s",
author = "Thakur, Nandan and
Zhang, Crystina and
Ma, Xueguang and
Lin, Jimmy",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-emnlp.481/",
pages = "9064--9083",
ISBN = "979-8-89176-335-7",
abstract = "Training robust retrieval and reranker models typically relies on large-scale retrieval datasets; for example, the BGE collection contains 1.6 million query-passage pairs sourced from various data sources.However, we find that certain datasets can negatively impact model effectiveness $\textemdash$pruning 8 out of 15 datasets from the BGE collection, reduces the training set size by 2.35$\times$, surprisingly increases nDCG@10 on BEIR by 1.0 point.This motivates a deeper examination of training data quality, with a particular focus on ``false negatives'', where relevant passages are incorrectly labeled as irrelevant.We utilize LLMs as a simple, cost-effective approach to *identify* and *relabel* false negatives in training datasets.Experimental results show that relabeling false negatives as true positives improves both E5 (base) and Qwen2.5-7B retrieval models by 0.7-1.4 points on BEIR and by 1.7-1.8 points at nDCG@10 on zero-shot AIR-Bench evaluation.Similar gains are observed for rerankers fine-tuned on the relabeled data, such as Qwen2.5-3B on BEIR.The reliability of LLMs to identify false negatives is supported by human annotation results. Our training dataset and code are publicly available."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="thakur-etal-2025-hard">
<titleInfo>
<title>Hard Negatives, Hard Lessons: Revisiting Training Data Quality for Robust Information Retrieval with LLMs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Nandan</namePart>
<namePart type="family">Thakur</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Crystina</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xueguang</namePart>
<namePart type="family">Ma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jimmy</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christos</namePart>
<namePart type="family">Christodoulopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carolyn</namePart>
<namePart type="family">Rose</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Violet</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-335-7</identifier>
</relatedItem>
<abstract>Training robust retrieval and reranker models typically relies on large-scale retrieval datasets; for example, the BGE collection contains 1.6 million query-passage pairs sourced from various data sources.However, we find that certain datasets can negatively impact model effectiveness —pruning 8 out of 15 datasets from the BGE collection, reduces the training set size by 2.35\times, surprisingly increases nDCG@10 on BEIR by 1.0 point.This motivates a deeper examination of training data quality, with a particular focus on “false negatives”, where relevant passages are incorrectly labeled as irrelevant.We utilize LLMs as a simple, cost-effective approach to *identify* and *relabel* false negatives in training datasets.Experimental results show that relabeling false negatives as true positives improves both E5 (base) and Qwen2.5-7B retrieval models by 0.7-1.4 points on BEIR and by 1.7-1.8 points at nDCG@10 on zero-shot AIR-Bench evaluation.Similar gains are observed for rerankers fine-tuned on the relabeled data, such as Qwen2.5-3B on BEIR.The reliability of LLMs to identify false negatives is supported by human annotation results. Our training dataset and code are publicly available.</abstract>
<identifier type="citekey">thakur-etal-2025-hard</identifier>
<location>
<url>https://aclanthology.org/2025.findings-emnlp.481/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>9064</start>
<end>9083</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Hard Negatives, Hard Lessons: Revisiting Training Data Quality for Robust Information Retrieval with LLMs
%A Thakur, Nandan
%A Zhang, Crystina
%A Ma, Xueguang
%A Lin, Jimmy
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Findings of the Association for Computational Linguistics: EMNLP 2025
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-335-7
%F thakur-etal-2025-hard
%X Training robust retrieval and reranker models typically relies on large-scale retrieval datasets; for example, the BGE collection contains 1.6 million query-passage pairs sourced from various data sources.However, we find that certain datasets can negatively impact model effectiveness —pruning 8 out of 15 datasets from the BGE collection, reduces the training set size by 2.35\times, surprisingly increases nDCG@10 on BEIR by 1.0 point.This motivates a deeper examination of training data quality, with a particular focus on “false negatives”, where relevant passages are incorrectly labeled as irrelevant.We utilize LLMs as a simple, cost-effective approach to *identify* and *relabel* false negatives in training datasets.Experimental results show that relabeling false negatives as true positives improves both E5 (base) and Qwen2.5-7B retrieval models by 0.7-1.4 points on BEIR and by 1.7-1.8 points at nDCG@10 on zero-shot AIR-Bench evaluation.Similar gains are observed for rerankers fine-tuned on the relabeled data, such as Qwen2.5-3B on BEIR.The reliability of LLMs to identify false negatives is supported by human annotation results. Our training dataset and code are publicly available.
%U https://aclanthology.org/2025.findings-emnlp.481/
%P 9064-9083
Markdown (Informal)
[Hard Negatives, Hard Lessons: Revisiting Training Data Quality for Robust Information Retrieval with LLMs](https://aclanthology.org/2025.findings-emnlp.481/) (Thakur et al., Findings 2025)
ACL