Closed mdrio closed 2 years ago
Hi,
We are working on it. We are trying to find out what is happening.
Álvaro
Ok, thanks.
Hi,
We found and solved the problem. There was a problem increasing the batch size with CuDNN. It will be fixed in the next release.
Thanks!
Using docker image dhealth/pylibs-toolkit:0.12.2-cudnn and a VGG network I encountered the following error:
Here is the code to reproduce the error:
Here is the log produced on my environment:
Note that the error does not occur if you swap the order of the tensors, i.e. you change the last two lines into:
No error occurs when using CPU or GPU without CUDNN.