Running those old configs today would be hard. They need pylearn2, pylearn2 needs an old version of theano, old versions of theano need old versions of CUDA, old versions of CUDA need old GPUs.

→ View original post on X — @goodfellow_ian,