HyperAI초신경

Visual Reasoning On Clevrer

평가 지표

Average-per ques.
Counterfactual-per opt.
Counterfactual-per ques.
Descriptive
Explanatory-per opt.
Explanatory-per ques.
Predictive-per opt.
Predictive-per ques.

평가 결과

이 벤치마크에서 각 모델의 성능 결과

비교 표
모델 이름Average-per ques.Counterfactual-per opt.Counterfactual-per ques.DescriptiveExplanatory-per opt.Explanatory-per ques.Predictive-per opt.Predictive-per ques.
모델 173.379.9650.8988.3789.1981.5684.8372.38
모델 273.179.649.7788.7989.1681.2484.9572.6
모델 367.5781.0151.0774.9890.8175.6282.968.61
모델 460.2566.6525.8981.3983.4272.7878.560.95
모델 588.0591.1274.8995.0498.1894.9893.1187.28
모델 675.5280.3846.5290.789.5882.8290.5282.03
모델 790.2494.8384.2993.496.391.9495.6891.35
모델 888.2791.4275.6194.0198.4795.9993.4987.48
모델 988.7191.2575.3594.7798.2595.4694.1689.25
모델 1069.6574.0542.2388.0887.6479.682.8668.7
think-before-you-simulate-symbolic-reasoning95.2496.6190.7296.4699.9499.8193.9693.96
모델 1291.1492.9780.0595.7698.8896.9895.6991.75
모델 1369.2178.0844.689.9595.9491.9874.7350.34