Adaptive Segment-level Reward: Bridging the Gap Between Action and Reward Space in Alignment

Re-assign community