Frustratingly Simple Retrieval Improves Challenging, Reasoning-Intensive Benchmarks

• URL: http://arxiv.org/abs/2507.01297v2
• Date: Sat, 05 Jul 2025 21:33:54 GMT
• Title: Frustratingly Simple Retrieval Improves Challenging, Reasoning-Intensive Benchmarks
• Abstract summary: Retrieval-augmented Generation (RAG) has primarily been studied in limited settings, such as factoid question answering.<n>In this work, we challenge this prevailing view on established, reasoning-intensive benchmarks: MMLU, MMLU Pro, AGI Eval, GPQA, and MATH.<n>We introduce CompactDS: a diverse, high-quality, web-scale datastore that achieves high retrieval accuracy and subsecond latency on a single-node.
• Score: 30.439565592162413
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Retrieval-augmented Generation (RAG) has primarily been studied in limited settings, such as factoid question answering; more challenging, reasoning-intensive benchmarks have seen limited success from minimal RAG. In this work, we challenge this prevailing view on established, reasoning-intensive benchmarks: MMLU, MMLU Pro, AGI Eval, GPQA, and MATH. We identify a key missing component in prior work: a usable, web-scale datastore aligned with the breadth of pretraining data. To this end, we introduce CompactDS: a diverse, high-quality, web-scale datastore that achieves high retrieval accuracy and subsecond latency on a single-node. The key insights are (1) most web content can be filtered out without sacrificing coverage, and a compact, high-quality subset is sufficient; and (2) combining in-memory approximate nearest neighbor (ANN) retrieval and on-disk exact search balances speed and recall. Using CompactDS, we show that a minimal RAG pipeline achieves consistent accuracy improvements across all benchmarks and model sizes (8B--70B), with relative gains of 10% on MMLU, 33% on MMLU Pro, 14% on GPQA, and 19% on MATH. No single data source suffices alone, highlighting the importance of diversity of sources (web crawls, curated math, academic papers, textbooks). Finally, we show that our carefully designed in-house datastore matches or outperforms web search engines such as Google Search, as well as recently proposed, complex agent-based RAG systems--all while maintaining simplicity, reproducibility, and self-containment. We release CompactDS and our retrieval pipeline, supporting future research exploring retrieval-based AI systems.

Related papers

• MM-Doc-R1: Training Agents for Long Document Visual Question Answering through Multi-turn Reinforcement Learning [74.07254720088926]
  We introduce MM-Doc-R1, a novel framework that employs an agentic, vision-aware workflow to address long document visual question answering.<n>We propose Similarity-based Policy Optimization (SPO), addressing baseline estimation bias in existing multi-turn reinforcement learning (RL) algorithms like GRPO.<n>Our experiments on the MMLongbench-Doc benchmark show that MM-Doc-R1 outperforms previous baselines by 10.4%.
  arXiv  Detail & Related papers  (2026-04-15T07:39:08Z)
• Progressive Searching for Retrieval in RAG [1.1912082737504726]
  Retrieval Augmented Generation (RAG) is a promising technique for mitigating two key limitations of large language models (LLMs): outdated information and hallucinations.<n>We propose a cost-effective searching algorithm for retrieval process.<n>Our findings demonstrate that progressive search in RAG systems achieves a balance between dimensionality, speed, and accuracy, enabling scalable and high-performance retrieval even for large databases.
  arXiv  Detail & Related papers  (2026-02-07T01:12:53Z)
• From HNSW to Information-Theoretic Binarization: Rethinking the Architecture of Scalable Vector Search [0.7804710977378487]
  This paper analyzes the architectural limitations of the dominant "HNSW + float32 + cosine similarity" stack.<n>We introduce and empirically evaluate an alternative information-theoretic architecture based on maximally informative binarization (MIB)<n>Results demonstrate retrieval quality comparable to full-precision systems, while achieving substantially lower latency and maintaining constant throughput at high request rates.
  arXiv  Detail & Related papers  (2025-12-16T23:24:37Z)
• MemSearcher: Training LLMs to Reason, Search and Manage Memory via End-to-End Reinforcement Learning [73.27233666920618]
  We propose MemSearcher, an agent workflow that iteratively maintains a compact memory and combines the current turn with it.<n>At each turn, MemSearcher fuses the user's question with the memory to generate reasoning traces, perform search actions, and update memory to retain only information essential for solving the task.<n>We introduce multi-context GRPO, an end-to-end RL framework that jointly optimize reasoning, search strategies, and memory management of MemSearcher Agents.
  arXiv  Detail & Related papers  (2025-11-04T18:27:39Z)
• MARAG-R1: Beyond Single Retriever via Reinforcement-Learned Multi-Tool Agentic Retrieval [50.30107119622642]
  Large Language Models (LLMs) excel at reasoning and generation but are inherently limited by static pretraining data.<n>Retrieval-Augmented Generation (RAG) addresses this issue by grounding LLMs in external knowledge.<n>MarAG-R1 is a reinforcement-learned multi-tool RAG framework that enables LLMs to dynamically coordinate multiple retrieval mechanisms.
  arXiv  Detail & Related papers  (2025-10-31T15:51:39Z)
• FrugalRAG: Learning to retrieve and reason for multi-hop QA [10.193015391271535]
  Large-scale fine-tuning is not needed to improve RAG metrics.<n>Supervised and RL-based fine-tuning can help RAG from the perspective of frugality.
  arXiv  Detail & Related papers  (2025-07-10T11:02:13Z)
• Benchmarking Deep Search over Heterogeneous Enterprise Data [73.55304268238474]
  We present a new benchmark for evaluating a form of retrieval-augmented generation (RAG)<n>RAG requires source-aware, multi-hop reasoning over diverse, sparsed, but related sources.<n>We build it using a synthetic data pipeline that simulates business across product planning, development, and support stages.
  arXiv  Detail & Related papers  (2025-06-29T08:34:59Z)
• MMSearch-R1: Incentivizing LMMs to Search [49.889749277236376]
  We present MMSearch-R1, the first end-to-end reinforcement learning framework that enables on-demand, multi-turn search in real-world Internet environments.<n>Our framework integrates both image and text search tools, allowing the model to reason about when and how to invoke them guided by an outcome-based reward with a search penalty.
  arXiv  Detail & Related papers  (2025-06-25T17:59:42Z)
• GeAR: Graph-enhanced Agent for Retrieval-augmented Generation [12.966494167631113]
  Retrieval-augmented Generation (RAG) relies on effective retrieval capabilities.<n>Traditional sparse and dense retrievers inherently struggle with multi-hop retrieval scenarios.<n>We introduce GeAR, a system that advances RAG performance through two key innovations.
  arXiv  Detail & Related papers  (2024-12-24T13:45:22Z)
• Blended RAG: Improving RAG (Retriever-Augmented Generation) Accuracy with Semantic Search and Hybrid Query-Based Retrievers [0.0]
  Retrieval-Augmented Generation (RAG) is a prevalent approach to infuse a private knowledge base of documents with Large Language Models (LLM) to build Generative Q&A (Question-Answering) systems. We propose the 'Blended RAG' method of leveraging semantic search techniques, such as Vector indexes and Sparse indexes, blended with hybrid query strategies. Our study achieves better retrieval results and sets new benchmarks for IR (Information Retrieval) datasets like NQ and TREC-COVID datasets.
  arXiv  Detail & Related papers  (2024-03-22T17:13:46Z)
• CFIR: Fast and Effective Long-Text To Image Retrieval for Large Corpora [3.166549403591528]
  This paper presents a two-stage Coarse-to-Fine Index-shared Retrieval (CFIR) framework, designed for fast and effective long-text to image retrieval. CFIR surpasses existing MLLMs by up to 11.06% in Recall@1000, while reducing training and retrieval times by 68.75% and 99.79%, respectively.
  arXiv  Detail & Related papers  (2024-02-23T11:47:16Z)
• SPRINT: A Unified Toolkit for Evaluating and Demystifying Zero-shot Neural Sparse Retrieval [92.27387459751309]
  We provide SPRINT, a unified Python toolkit for evaluating neural sparse retrieval. We establish strong and reproducible zero-shot sparse retrieval baselines across the well-acknowledged benchmark, BEIR. We show that SPLADEv2 produces sparse representations with a majority of tokens outside of the original query and document.
  arXiv  Detail & Related papers  (2023-07-19T22:48:02Z)
• Injecting Domain Adaptation with Learning-to-hash for Effective and Efficient Zero-shot Dense Retrieval [49.98615945702959]
  We evaluate LTH and vector compression techniques for improving the downstream zero-shot retrieval accuracy of the TAS-B dense retriever. Our results demonstrate that, unlike prior work, LTH strategies when applied naively can underperform the zero-shot TAS-B dense retriever on average by up to 14% nDCG@10.
  arXiv  Detail & Related papers  (2022-05-23T17:53:44Z)
• IRLI: Iterative Re-partitioning for Learning to Index [104.72641345738425]
  Methods have to trade between obtaining high accuracy while maintaining load balance and scalability in distributed settings. We propose a novel approach called IRLI, which iteratively partitions the items by learning the relevant buckets directly from the query-item relevance data. We mathematically show that IRLI retrieves the correct item with high probability under very natural assumptions and provides superior load balancing.
  arXiv  Detail & Related papers  (2021-03-17T23:13:25Z)
• RadixSpline: A Single-Pass Learned Index [84.84747738666263]
  We introduce RadixSpline (RS), a learned index that can be built in a single pass over the data. RS achieves competitive results on all datasets, despite the fact that it only has two parameters.
  arXiv  Detail & Related papers  (2020-04-30T01:56:54Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.