Fetching the paper…

Adapt2Reward: Adapting Video-Language Models to Generalizable Robotic Rewards via Failure Prompts · Around