HyperAI超神经

Visual Reasoning On Phyre 1B Cross

评估指标

AUCCESS

评测结果

各个模型在此基准测试上的表现结果

比较表格
模型名称AUCCESS
learning-long-term-visual-dynamics-with42.2
forward-prediction-for-physical-reasoning40.3
physical-reasoning-using-dynamics-aware39.9
phyre-a-new-benchmark-for-physical-reasoning36.8