When I run this program has encountered a problem,Can you help me？ about kaggle-galaxies

[benanne]

That looks pretty bad - maybe you just don't have enough memory though. How much memory does your card have? You'll probably need at least 2GB to run the code.

[lixunlove]

Thank you for your reply .
My card is GeForce GTX 650 with 2GB memory ,
Under what circumstances would cause “ValueError: need at least one array to concatenate”?

[benanne]

I think the problem is the 'initialization error', not the concatenation error. That's probably just a consequence of the previous error.

Another possibility is that it is unrelated and that the dataset could not be loaded. Did you download the JPEG files and extract them in the right locations?

[lixunlove]

Wether I copy the train and test images to /dev/shm by running python copy data to shm.py or modify the from_ram parameter to False is the same error.
PS:
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2014 NVIDIA Corporation
Built on Thu_Jul_17_21:41:27_CDT_2014
Cuda compilation tools, release 6.5, V6.5.12

[benanne]

Okay. Assuming both errors are related: try export CUDA_LAUNCH_BLOCKING=1 and then running your code (see http://comments.gmane.org/gmane.comp.mathematics.theano.user/467 ). Maybe that will give you a better error message.

If that doesn't help, I would suggest asking on the Theano mailinglist as this is probably either an issue with Theano or a hardware issue, and not really a problem with this code.