haithanhp

Hi @Catosine, Thanks for your response. I use train_search.py and train_imagenet.py in cnn folder and did not change anything about batch_size, learning rate, and so on. Btw, I have one...

Hi @Catosine, Thanks for your response. Sure. I did try the code you sent and can reproduce the ImageNet accuracy. Thanks for your suggestions.

@NdaAzr , Just get your searched architecture and put a name in genotypes.py (My_Model= [your architecture searched]). For example: DARTS_V3 = Genotype(normal=[('skip_connect', 0), ('sep_conv_3x3', 1), ('skip_connect', 0), ('skip_connect', 1), ('sep_conv_3x3',...

Thanks for your information. Do you know why the models in pytorch are very huge (**48.2MB**)? I see same scale models (Mobilenet v1, v2) with **4.4M** and **3.9M** (#parameters) only...

Thanks for your answer. I have an example of one part of DenseNet-40 (k=12): module.features.init_conv.weight : torch.Size([24, 3, 3, 3]) module.features.denseblock_1.dense_basicblock_1.conv_33.norm.weight : torch.Size([24]) module.features.denseblock_1.dense_basicblock_1.conv_33.norm.bias : torch.Size([24]) module.features.denseblock_1.dense_basicblock_1.conv_33.norm.running_mean : torch.Size([24]) module.features.denseblock_1.dense_basicblock_1.conv_33.norm.running_var...

1. When the second dimension of conv.weight is pruned to 26 (prune away 10), the dimension of input activation is still 36 and it won't be matched. How can you...

Yes, I see. Also, do you public the code for DenseNet and Resnet experiments? I also need to reproduce all your experiments for evaluation. Thanks.