Research map / Large language models / Agents, code and math
JudgeBoard: Benchmarking and Enhancing Small Language Models for Reasoning Evaluation
Zhenyu Bi, Gaurav Srivastava, Yang Li, Swastik Roy, Meng Lu, Morteza Ziyadi, Xuan Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use multi-agent profiled debate (MAJ) instead of single judge | Qwen3-14B · MATH Algebra | overall judging accuracy: 0.6684 -> 0.7789 | Improved |
| use MAJ instead of single judge | Qwen3-14B · MATH Number Theory | overall judging accuracy: 0.6923 -> 0.9385 | Improved |
| use MAJ instead of single judge | Qwen3-14B · MATH Counting and Probability | overall judging accuracy: 0.7111 -> 0.9056 | Improved |
| use MAJ instead of single judge | Qwen3-8B · MATH Number Theory | overall judging accuracy: 0.6308 -> 0.9 | Improved |
| use MAJ instead of single judge | Qwen3-4B · MATH Algebra | overall judging accuracy: 0.6421 -> 0.7632 | Improved |
| use MAJ instead of single judge | Gemma3-12b · MATH Algebra | overall judging accuracy: 0.6737 -> 0.7 | Improved |
| use MAJ instead of single judge | Gemma3-12b · MATH Number Theory | overall judging accuracy: 0.7923 -> 0.7231 | Hurt |
| use MAJ instead of single judge | Phi4 · MATH Algebra | overall judging accuracy: 0.6737 -> 0.6421 | Hurt |
| use MAJ instead of single judge | Phi4 · MATH Number Theory | overall judging accuracy: 0.7077 -> 0.6692 | Hurt |
| enable reasoning mode (ablation) | Qwen3-4B · MATH Algebra | overall judging accuracy: 0.6421 -> 0.6474 | Improved |
| enable reasoning mode (ablation) | Qwen3-4B · MATH Number Theory | overall judging accuracy: 0.7384 -> 0.7000 | Hurt |
| enable reasoning mode (ablation) | Qwen3-14B · MATH Number Theory | overall judging accuracy: 0.6923 -> 0.7077 | Improved |
Closest papers by meaning
- Evaluating Judges as Evaluators: The JETTS Benchmark of LLM-as-Judges as Test-Time Scaling Evaluators (ICML 2025 · similarity 0.91)
- JudgeBench: A Benchmark for Evaluating LLM-Based Judges (ICLR 2025 · similarity 0.90)
- MR-GSM8K: A Meta-Reasoning Benchmark for Large Language Model Evaluation (ICLR 2025 · similarity 0.90)
- JudgeLM: Fine-tuned Large Language Models are Scalable Judges (ICLR 2025 · similarity 0.90)
- Sample Smart, Not Hard: Correctness-First Decoding for Better Reasoning in LLMs (ICLR 2026 · similarity 0.90)
- metabench - A Sparse Benchmark of Reasoning and Knowledge in Large Language Models (ICLR 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
