Closed yian2204 closed 2 days ago
The training settings for CIFAR10 and CIFAR100 are exactly the same. Have you tried to experiment on CIFAR10? Or have you tried to train on other models such as XNOR and Bi-real? You can also send me an email to discuss more details: ruichen.ma@std.uestc.edu.cn.
Let me contact you by email.
Can I know the specific experimental setup of your A&B cifar100? I currently reproduce your code, but it only results at 0.677. Or can you tell me why there is this gap?