Theano error while training

[manishprasadthapliyal]

On running the train module from this github repository

import train
train.trainer(list)

While running the above code. we are getting error in Theano. error file has been attached.

[error.txt](https://github.com/jatana-research/email-summarization/files/2414610/error.txt)
error2.txt

[kushalchauhan98]

You told me that you got this error when running the code on CPU here. The code should work fine on CPU. I wouldn't know why is CuDnn is even appearing in the error traceback when using CPU. I cannot reproduce the problem as I am using Ubuntu, and it works fine here. Can you try using
device=gpu
instead of
device=cuda0
in the .theanorc and try running it on GPU again?

[manishprasadthapliyal]

Below is the .theanorc file we are using.

[global]
device = gpu
floatX = float32

[cuda]
root = /usr/local/cuda