lisa-lab/pylearn2

reproducing maxout paper results on mnist

Opened this issue · 0 comments

kmyi commented

There's batch size missing on mnist_continued.py
I am not sure, but adding

    batch_size: 128,

in algorithm seems to fix it.