# Projecting BrowseComp-Plus onto ClimbMix: Toward More Realistic Corpora for Agentic Search

arXiv: https://arxiv.org/abs/2608.20317 · HTML: https://mlsanity.com/paper/2608.20317/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Web, search, and research agents, one of 16 directions in LLM agents and tool use.
- Builds on 24 of its 31 references that are in the map.
- First posted on arXiv 20 Aug 2026.

## Placed in

- LLM agents and tool use / [Web, search, and research agents](https://mlsanity.com/agents/web-search-and-research-agents/)

## Builds on (most cited first, 24 of 24)

- [MS MARCO: A Human Generated MAchine Reading COmprehension Dataset](https://mlsanity.com/paper/1611.09268/) (2016)
- [The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale](https://mlsanity.com/paper/2406.17557/) (2024)
- [BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents](https://mlsanity.com/paper/2504.12516/) (2025)
- [Fact, Fetch, and Reason: A Unified Evaluation of Retrieval-Augmented Generation](https://mlsanity.com/paper/2409.12941/) (2024)
- [Nemotron-CC: Transforming Common Crawl into a Refined Long-Horizon Pretraining Dataset](https://mlsanity.com/paper/2412.02595/) (2024)
- [Nemotron-CLIMB: CLustering-based Iterative Data Mixture Bootstrapping for Language Model Pre-training](https://mlsanity.com/paper/2504.13161/) (2025)
- [DeepResearchGym: A Free, Transparent, and Reproducible Evaluation Sandbox for Deep Research](https://mlsanity.com/paper/2505.19253/) (2025)
- [DeepSearchQA: Bridging the Comprehensiveness Gap for Deep Research Agents](https://mlsanity.com/paper/2601.20975/) (2026)
- [Beyond Semantic Similarity: Rethinking Retrieval for Agentic Search via Direct Corpus Interaction](https://mlsanity.com/paper/2605.05242/) (2026)
- [OpenResearcher: A Fully Open Pipeline for Long-Horizon Deep Research Trajectory Synthesis](https://mlsanity.com/paper/2603.20278/) (2026)
- [Ragnar"ok: A Reusable RAG Framework and Baselines for TREC 2024 Retrieval-Augmented Generation Track](https://mlsanity.com/paper/2406.16828/) (2024)
- [MoNaCo: More Natural and Complex Questions for Reasoning Across Dozens of Documents](https://mlsanity.com/paper/2508.11133/) (2025)
- [Is Grep All You Need? How Agent Harnesses Reshape Agentic Search](https://mlsanity.com/paper/2605.15184/) (2026)
- [Search-Time Data Contamination](https://mlsanity.com/paper/2508.13180/) (2025)
- [Rethinking Agentic Search with Pi-Serini: Is Lexical Retrieval Sufficient?](https://mlsanity.com/paper/2605.10848/) (2026)
- [GrepSeek: Training Search Agents for Direct Corpus Interaction](https://mlsanity.com/paper/2605.29307/) (2026)
- [Towards Retrieving Interaction Spaces for Agentic Search](https://mlsanity.com/paper/2606.06880/) (2026)
- [Total Recall QA: A Verifiable Evaluation Suite for Deep Research Agents](https://mlsanity.com/paper/2603.18516/) (2026)
- [A New Role for Relevance: Guiding Corpus Interaction in Agentic Search](https://mlsanity.com/paper/2607.24223/) (2026)
- [Search, Inspect, Fetch: Exploiting Structure-Aware Boolean Retrieval for Deep-Search Agents](https://mlsanity.com/paper/2608.02751/) (2026)
- [NanoKnow: How to Know What Your Language Model Knows](https://mlsanity.com/paper/2602.20122/) (2026)
- [ORBIT: Scalable and Verifiable Data Generation for Search Agents on a Tight Budget](https://mlsanity.com/paper/2604.01195/) (2026)
- [Superintelligent Retrieval Agent: The Next Frontier of Agentic Retrieval](https://mlsanity.com/paper/2605.06647/) (2026)
- [Towards Query-Agnostic RAG Evaluation via Query Coverage and Claim Verifiability](https://mlsanity.com/paper/2608.11238/) (2026)

## Abstract (arXiv)

The BrowseComp-Plus benchmark disentangled the evaluation of agentic search by replacing opaque web search with a fixed corpus, so that an agent's role can be separated from the retriever's. That corpus, however, holds only about 100K documents and was assembled from the supporting documents of the benchmark's own queries plus mined hard negatives, so the evidence and the distractors were both selected per query. We introduce BrowseComp-PlusCM, which keeps the BrowseComp-Plus questions but relocates their evidence to ClimbMix, a 400B-token, 553M-document mixture of web text released by NVIDIA for pre-training language models and built without reference to any benchmark. Our main contribution is the projection pipeline that makes this possible: it decomposes each question into atomic reasoning hops and grounds every hop in the new corpus, retaining a question only when automatic verification, an independent agent, and human review all confirm that every hop is supported. The pipeline is dataset-agnostic and applies to any benchmark whose questions decompose into verifiable facts. Applied to the 830 BrowseComp-Plus test questions, our pipeline yields 57 fully grounded questions with question-level relevance judgments. Projection shifts the difficulty onto retrieval, as the strongest agent we evaluate loses five points of answer accuracy but sees its evidence recall fall from 84.3% to 21.4% while issuing 63% more search calls. As the first of a series of projections, we release the pipeline, the benchmark, and our analyses at https://github.com/castorini/cmass.