PRISM: Pushing the Frontier of Deep Think via Process Reward Model-Guided Inference.

arXiv:2603.02479v2 Announce Type: replace Abstract: DEEPTHINK methods improve reasoning by generating, refining, and aggregating populations of candidate solutions, which enables strong performance on complex mathematical and scientific tasks. However, existin