arXiv AI By Zihan Dong, Zhixian Zhang, Yang Zhou, Can Jin, Ruijia Wu, Linjun Zhang

Evaluating LLMs When They Do Not Know the Answer: Statistical Evaluation of Mathematical Reasoning via Comparative Signals

Read the original on arXiv AI →

arXiv:2602. 03061v2 Announce Type: replace-cross Abstract: Evaluating mathematical reasoning in LLMs is constrained by limited benchmark sizes and inherent model stochasticity, yielding high-variance accuracy estimates and unstable rankings across platforms.

Summary generated by The Flow from the publisher's feed. The full article lives at arXiv AI.

arXiv AI
Jul 14

PiCSAR: Probabilistic Confidence Selection And Ranking for Reasoning Chains

arXiv:2508. 21787v3 Announce Type: replace-cross Abstract: Best-of-n sampling improves the accuracy of large language models (LLMs) and large reasoning models (LRMs) by generating multiple candidate solutions and selecting the one with the highest reward.

By Joshua Ong Jun Leang, Zheng Zhao, Aryo Pradipta Gema, Sohee Yang, Wai-Chung Kwan, Xuanli He, Wenda Li, Pasquale Minervini, Eleonora Giunchiglia, Shay B. Cohen
arXiv AI
Jun 2

ReasonBENCH: Benchmarking the (In)Stability of LLM Reasoning

arXiv:2512. 07795v2 Announce Type: replace Abstract: Benchmark scores for LLM reasoning systems are reported as single numbers, yet the same model, strategy, and task can produce meaningfully different answers and costs across repeated executions, even under greedy decoding (T = 0).

By Nearchos Potamitis, Vansh Ramani, Har Ashish Arora, Dhairya Kuchhal, Lars Klein, Akhil Arora