UC Berkeley 团队论文 SuperMimic 获 CORL 2026 最佳会议论文奖

Model-based training could cut the compute and sim-hours that RL pipelines demand, but the sim-to-real gap on contact-rich edge cases becomes the new bottleneck. Teams building motion-tracking skills for humanoids can test SuperMimic against their existing DreamerV3-style or PPO pipelines using the open GitHub code. Anyone deploying G1-class robots in unstructured environments should weigh the noted failure modes on collisions and slippage before relying on it.

Gorden Sun@Gorden_SunSource

Evaluation shifts from comparing static code screenshots to judging how agents actually behave inside an app, which changes how generation quality can be measured for web development agents. Builders of coding agents and eval pipelines can use the two open-sourced trace datasets to train reward models or benchmark correctness without collecting their own user complaints data. Compared with human-preference-only arenas, this adds an automated signal that scales with computer-use capability.

lmarena.ai@lmarena_aiSource

The core takeaway is that evaluation signal can be harvested from real user interactions instead of hand-written test suites, shifting agent benchmarking toward live platform data. Researchers and teams building computer-use agents can apply this pattern to score traces with LLM judges or reward models trained on user complaints. Code Arena and Agent Arena already supply the battle and complaint data, giving practitioners a concrete starting point for evaluation pipelines.

lmarena.ai@lmarena_aiSource