@@ -87,6 +87,7 @@ def build_lookahead(*parameters, **kwargs):
8787 (build_lookahead , {'lr' : 1e-2 , 'weight_decay' : 1e-3 }, 200 ),
8888 (AdaBelief , {'lr' : 1e-2 , 'weight_decay' : 1e-3 }, 200 ),
8989 (AdaBound , {'lr' : 1e-2 , 'gamma' : 0.1 , 'weight_decay' : 1e-3 }, 200 ),
90+ (AdaBound , {'lr' : 1e-2 , 'gamma' : 0.1 , 'weight_decay' : 1e-3 , 'amsbound' : True }, 200 ),
9091 (AdamP , {'lr' : 1e-3 , 'weight_decay' : 1e-3 }, 800 ),
9192 (DiffGrad , {'lr' : 1e-2 , 'weight_decay' : 1e-3 }, 200 ),
9293 (DiffRGrad , {'lr' : 1e-1 , 'weight_decay' : 1e-3 }, 200 ),
@@ -103,6 +104,7 @@ def build_lookahead(*parameters, **kwargs):
103104 (build_lookahead , {'lr' : 5e-1 , 'weight_decay' : 1e-3 }, 500 ),
104105 (AdaBelief , {'lr' : 5e-1 , 'weight_decay' : 1e-3 }, 200 ),
105106 (AdaBound , {'lr' : 5e-1 , 'gamma' : 0.1 , 'weight_decay' : 1e-3 }, 200 ),
107+ (AdaBound , {'lr' : 1e-2 , 'gamma' : 0.1 , 'weight_decay' : 1e-3 , 'amsbound' : True }, 200 ),
106108 (AdamP , {'lr' : 5e-1 , 'weight_decay' : 1e-3 }, 500 ),
107109 (DiffGrad , {'lr' : 15 - 1 , 'weight_decay' : 1e-3 }, 500 ),
108110 (DiffRGrad , {'lr' : 1e-1 , 'weight_decay' : 1e-3 }, 200 ),
0 commit comments