Jeshua Lacock
Jeshua Lacock
I would if I knew it. I found an AWS image configured for Torch and archived and copied it to my machine, then ran the install script.
Nice, thanks, I didn't know that! Most recent: `commit c9b29cf41ec714ee45b4799c2bd76e82d1b1f267` `Author: soumith ` `Date: Wed Dec 21 07:25:57 2016 -0800`
Apparently it has something to do the latest torch normalization. Its definitely not a saturation or swapped channel problem.
Do you have libcudnn.so.5 installed to a path defined in LD_LIBRARY_PATH? Or alternatively, is it installed to path defined in the CUDNN_PATH env var?
You have installed cuDNN 7.1, but only cuDNN 5.x is supported. If you want those use fast neural style without installing cuDNN 5.1 just use this flag: `-use_cudnn -1`
p.s. the training I was attempting when it ran out of memory should have taken around 3GB and the card has 6GB. I also tried setting the batch_size to 1...
Did you base it on Torch 7 for iOS? Did you get permission from @jcjohnson to use this project for commercial use? He never responded to my request for permission...
@sheerun Come on now - 3 identical posts for your advertisement?
I think it is in poor taste and not really appreciated, personally speaking.
I wonder what percent you plan on donating to jcjohnson? I suspect he has more time invested in your "solution" than you do.