{"schema":"https://assignee.net/schemas/benchmark-evidence-v1","schema_version":"1.0","contract_version":"benchmark-evidence-v1.0","contract_updated":"2026-06-01","schema_documentation":"https://assignee.net/schemas","changelog_url":"https://assignee.net/changelog","publisher":{"name":"Assignee Research","url":"https://assignee.net"},"html_url":"https://assignee.net/benchmarks/evidence?model=Qwen2.5&bench=MATH","json_url":"https://assignee.net/benchmarks/evidence.json?model=Qwen2.5&bench=MATH","model":"Qwen2.5","benchmark":"MATH","source_count":9,"source_coverage":{"record_count":9,"distinct_source_count":9,"coverage_level":"BROAD","basis":"distinct public paper URLs or titles in this evidence cluster"},"source_profile":{"source_url_count":9,"missing_source_url_count":0,"domains":["arxiv.org","doi.org"],"year_min":2024,"year_max":2026,"basis":"public source URLs, source titles, and reported publication years in this evidence cluster"},"reported_range":{"min_score_pct":27.0,"max_score_pct":81.0},"spread_pp":54.0,"severity":"HIGH","entries":[{"model":"Qwen2.5","benchmark":"MATH","score_pct":81.0,"source_title":"LANPO: Bootstrapping Language and Numerical Feedback for Reinforcement Learning in LLMs","source_url":"http://arxiv.org/abs/2510.16552v1","source_domain":"arxiv.org","year":2025},{"model":"Qwen2.5","benchmark":"MATH","score_pct":75.0,"source_title":"Do LLMs Overthink Basic Math Reasoning? Benchmarking the Accuracy-Efficiency Tradeoff in Language Models","source_url":"http://arxiv.org/abs/2507.04023v3","source_domain":"arxiv.org","year":2025},{"model":"Qwen2.5","benchmark":"MATH","score_pct":71.3,"source_title":"Benchmarking LLMs' Mathematical Reasoning with Unseen Random Variables Questions","source_url":"http://arxiv.org/abs/2501.11790v4","source_domain":"arxiv.org","year":2025},{"model":"Qwen2.5","benchmark":"MATH","score_pct":62.1,"source_title":"Qwen2.5 Technical Report","source_url":"http://arxiv.org/abs/2412.15115v2","source_domain":"arxiv.org","year":2024},{"model":"Qwen2.5","benchmark":"MATH","score_pct":46.4,"source_title":"MINT-CoT: Enabling Interleaved Visual Tokens in Mathematical Chain-of-Thought Reasoning","source_url":"http://arxiv.org/abs/2506.05331v1","source_domain":"arxiv.org","year":2025},{"model":"Qwen2.5","benchmark":"MATH","score_pct":45.67,"source_title":"MathMixup: Boosting LLM Mathematical Reasoning with Difficulty-Controllable Data Synthesis and Curriculum Learning","source_url":"http://arxiv.org/abs/2601.17006v1","source_domain":"arxiv.org","year":2026},{"model":"Qwen2.5","benchmark":"MATH","score_pct":44.1,"source_title":"LLaDA-MoE: A Sparse MoE Diffusion Language Model","source_url":"http://arxiv.org/abs/2509.24389v1","source_domain":"arxiv.org","year":2025},{"model":"Qwen2.5","benchmark":"MATH","score_pct":34.8,"source_title":"SwS: Self-aware Weakness-driven Problem Synthesis in Reinforcement Learning for LLM Reasoning","source_url":"http://arxiv.org/abs/2506.08989v1","source_domain":"arxiv.org","year":2025},{"model":"Qwen2.5","benchmark":"MATH","score_pct":27.0,"source_title":"s1: Simple test-time scaling","source_url":"https://doi.org/10.48550/arxiv.2501.19393","source_domain":"doi.org","year":2025}],"interpretation":"This record groups score claims extracted from papers for the same model and benchmark label. A nonzero spread means the public literature reports different values for this cluster.","limitations":["Differences are not automatically errors.","Reported values may differ because of prompts, dataset versions, evaluation protocols, scoring rules, preprocessing, fine-tuning, or reporting conventions.","Source papers remain authoritative for their own claims."]}