PACE-Bench

PACE-Bench, developed by Hugging Face, is a benchmark that evaluates self-evolving agents on their ability to adapt physics designs to hidden environmental shifts. It uniquely tests continual adaptation capabilities beyond static problem-solving.

Two-hop subgraph: this entity, every entity it directly relates to, and every entity those neighbors relate to. Drag a node, scroll to zoom, click to inspect — or click any neighbor and re-center the atlas there.