Sirius083

I have the same gpu on windows and run for densenet(L=40, k=12), the total training time is 5h10minutes. (without imgaug.MapImage(lambda x: x - pp_mean) and PrefetchData) But other denseNet implementation...

I have the same question one this issue. Since resnet-32 on cifar10 takes 1/3 of training time compared with densenet. Do you know the problem? thanks in advance.

@shirleychangyuanyuan Hello, I have the same question as you. Did you find the answer now? Thanks in advance

> @Sirius083 or @joeyearsley can you help me? I use another effcient densenet implementation at https://github.com/cybertronai/gradient-checkpointing easy to implement, just add a few lines at the begining of your code.

> @Sirius083 Did you see that your gpu memory went down and training time per second went up when you used yaroslav's memory_saving_gradients? Also, are you using Windows or Linux?...

@gitrdonator sorry I did not use keras, I use tensorflow(1.9.0 on windows) and python 3.6 I just add these lines before the model defination and training part. I think the...

@gitrdonator I just add the few lines before in this code in cifar10-densenet.py https://github.com/YixuanLi/densenet-tensorflow I did not perticular add anything else , which means I add gradient checkpoiting to all...

@gitrdonator I said it works on tensorflow 1.9 on windows , I never tried it on tensorflow 1.5.

you can (1) generate .pb file from checkpoint file ```python def generate_pb_file(model_dir,output_node_name,option = 'latest'): ''' parameters: model_dir: the output model directory input_node_name: input name in graph output_node_name: output node name...

Hi, have you solved the problem? I got similar trend for validation cost on cifar10 and cifar100(increase as learning rate decrease)