lang tang

icon company xmu icon location 深圳 icon location working@Tencent now, focus on deep learning inference acceleration, want to build some interesting AI products wechat: frankktang96

Results 4 comments of lang tang

I don't think this is a practicable api for GPU usage grows with batchsize ...

When I use Tesla V100 gpu, batchsize=256 with input size 3x224x224 is ok. If you have less than 32g gpu memory then you can consider reduce the batchsize or input...

Hello! Could you please tell me how to do the cutout operation?

I implement this in my neural network, you can add cutout op at the beginning of the forward...

CIFAR_INPUT_ITER类_init_函数中np.save/np.load的作用

这里是为了让数据能连续存储，dali要求数据的存储是连续的来增加locality，提高访问速度

Use dali to train tinyimagenet does not reduce iterate time

Hi, your iterate time seems unstable, is your gpu running other applications?