Vector Policy Optimization: Training for Diversity Improves Test-Time Search

by [unverified] · free · Last verified 2026-06-17T08:42:23.289Z

Language models must now generalize out of the box to novel environments and work inside inference-scaling search procedures, such as AlphaEvolve, that select rollouts with a variety of task-specific reward functions. Unfortunately, the standard paradigm of LLM post-training optimizes a pre-speci...