3 papers
cs.LG2026
MaxProof: Scaling Mathematical Proof with Generative-Verifier RL and Population-Level Test-Time Scaling
Jiacheng Chen, Xinyu Zhang, Shunkai Zhang +20
We present MaxProof, a population-level test-time scaling framework for competition-level mathematical proof in the MiniMax-M3 series. M3 first trains three proof-oriented capabili…
cs.CL2026
Temporal Leakage in Search-Engine Date-Filtered Web Retrieval: A Retrospective Forecasting Case Study
Ali El Lahib, Ying-Jieh Xia, Zehan Li +2
Search-engine date filters are widely used to enforce pre-cutoff retrieval in retrospective evaluations of search-augmented forecasters. We show this approach is unreliable across…
cs.CL2026
Simulated Ignorance Fails: A Systematic Study of LLM Behaviors on Forecasting Problems Before Model Knowledge Cutoff
Zehan Li, Yuxuan Wang, Ali El Lahib +2
Evaluating LLM forecasting capabilities is constrained by a fundamental tension: prospective evaluation offers methodological rigor but prohibitive latency, while retrospective for…