vlawhern / arl-eegmodels

This is the Army Research Laboratory (ARL) EEGModels Project: A Collection of Convolutional Neural Network (CNN) models for EEG signal classification, using Keras and Tensorflow
Other
1.14k stars 284 forks source link

InvalidArgumentError: Default AvgPoolingOp only supports NHWC on device type CPU #14

Closed jgcaffari1 closed 4 years ago

jgcaffari1 commented 4 years ago

When I try to fit the eegnet model, I get the following error:

InvalidArgumentError: Default AvgPoolingOp only supports NHWC on device type CPU [[{{node average_pooling2d/AvgPool}} = AvgPoolT=DT_FLOAT, _class=["loc:@training/Adam/gradients/average_pooling2d/AvgPool_grad/AvgPoolGrad"], data_format="NCHW", ksize=[1, 1, 1, 4], padding="VALID", strides=[1, 1, 1, 4], _device="/job:localhost/replica:0/task:0/device:CPU:0"]]

I have cuda installed, but it does not appear to be working. What is the appropriate cuda setup for gpu support?

vlawhern commented 4 years ago

Yeah that error comes up if you have tensorflow-cpu installed (or I'm guessing a broken tensorflow-gpu install).

Assuming you have a valid CUDA install, you should just be able to use pip to install tensorflow-gpu; i.e.:

pip install tensorflow-gpu==1.X or pip install tensorflow-gpu==2.0.0

since this code works with Tensorflow 2 as well. Note that I do all my work in Linux; I might not be able to help much if you're using Windows and CUDA..

jgcaffari1 commented 4 years ago

This fixed my problem. Thank you!

On Mon, Nov 11, 2019 at 6:11 PM vlawhern notifications@github.com wrote:

Yeah that error comes up if you have tensorflow-cpu installed (or I'm guessing a broken tensorflow-gpu install).

Assuming you have a valid CUDA install, you should just be able to use pip to install tensorflow-gpu; i.e.:

pip install tensorflow-gpu==1.X or pip install tensorflow-gpu==2.0.0

since this code works with Tensorflow 2 as well. Note that I do all my work in Linux; I might not be able to help much if you're using Windows and CUDA..

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/vlawhern/arl-eegmodels/issues/14?email_source=notifications&email_token=AFI4D7JSKCQNY2Q3BVWZ7ZLQTHYC3A5CNFSM4JL4G2JKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEDYSO7Y#issuecomment-552675199, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFI4D7NSMBP74XD2PTYZWDDQTHYC3ANCNFSM4JL4G2JA .

vlawhern commented 4 years ago

Good to hear! Closing this but feel free to reopen if there are any other issues.