Fix one problem related to strides by copying the strides infos
from the input to the output.

Added __eq__ and __hash__ methods.
Fixed c_code_cache_version
Check if axis=None before bailing out.

Fix optimization not always registered and python 3 crash

Moved general FAQ to top-level index page.

[BUG] Important bug fix in an optimization.

[WIP] Fast compile gpu

Propagate shape information to ConvOp gradients

Abergeron dnn