ImageNet pre-trained EfficientNet Lite models.
eval reaults:
efficientnet_lite0
TEST Iter 0: loss = 2.100231, Top-1 err = 0.282700, Top-5 err = 0.098280, val_time = 120.648957
efficientnet_lite1
TEST Iter 0: loss = 2.076898, Top-1 err = 0.252940, Top-5 err = 0.079880, val_time = 126.869352
efficientnet_lite2
TEST Iter 0: loss = 1.929238, Top-1 err = 0.228660, Top-5 err = 0.064640, val_time = 142.668548
efficientnet_lite3
TEST Iter 0: loss = 1.782202, Top-1 err = 0.210920, Top-5 err = 0.056260, val_time = 147.359098
efficientnet_lite4
TEST Iter 0: loss = 1.714834, Top-1 err = 0.196580, Top-5 err = 0.049440, val_time = 158.336004