These are solutions to the exercises up at the Stanford OpenClassroom Deep Learning class and Andrew Ng's UFLDL Tutorial. I looked around for copies of the solutions so I could compare notes, but almost everything I found was incomplete or obvious wrong. I don't promise that these don't have bugs, but they at least give outputs within the range of the expected outputs for the assignments.
Apologies for the mess. I'll clean this up when I have some spare time. Pull requests welcome, of course.