Paper Space

I just started using paperspace. Setup was as easy as Jeremy said - thanks!

As a test of my machine, I wanted to use the fast-style-transfer code here (which is a fun and easy network to try):
https://github.com/lengstrom/fast-style-transfer

I was able to get evalute.py to work on my local machine, but when I try and run it on my paperspace machine I get this error:

2018-02-08 09:17:49.635963: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2
terminate called after throwing an instance of 'std::bad_alloc'
  what():  std::bad_alloc
Aborted (core dumped)

Which seems to imply I am running out of space? Any thoughts on how I might be going about this wrong?

This usually works in just a couple seconds on my local machine - but before I get this error on paperspace, it waits at least 45 seconds. Any thoughts on what I might investigate would be much appreciated.

Thanks.

Hey, in case anyone else comes across this problem:

I found that the image I was trying to use as the source of the style transfer was too big in pixel size (it was over 3000x2000). I reduced my image to something smaller (like 1024x768) and the error went away.

Hope this helps someone!

1 Like