GitHub

So, basically we need to implement most of what's in torch7-libv2 here. Let's start off with this list:

Ordered:

Not needed:

2017-06-17: SpatialDropoutX21.lua So, the reason I needed SpatialDropoutX21 was because luatorch's SpatialDropout didn't premultiply by (1/(1-p)) pytorch's nn.Dropout2d does premultiply... So, this is not needed anymore.
2017-06-18: Random.lua -- Let's see how far we can get without implementing these.
2017-06-18: Zero.lua

Name		Name	Last commit message	Last commit date
Latest commit History 74 Commits
tests		tests
Makefile		Makefile
README.md		README.md
__init__.py		__init__.py
addgaussian.py		addgaussian.py
alexnet.py		alexnet.py
attention.py		attention.py
binarystochastic.py		binarystochastic.py
circumventablebatchnorm1d.py		circumventablebatchnorm1d.py
crc32c.py		crc32c.py
debug.py		debug.py
device.py		device.py
dropoutgraph.py		dropoutgraph.py
event_reader.py		event_reader.py
highwaylayer.py		highwaylayer.py
highwaylstmlayer.py		highwaylstmlayer.py
highwaynormlayer.py		highwaynormlayer.py
identity.py		identity.py
imageutil.py		imageutil.py
layernormlstm.py		layernormlstm.py
linearnorm.py		linearnorm.py
linearpure.py		linearpure.py
normalize.py		normalize.py
nullable.py		nullable.py
random_source.py		random_source.py
randomlayer.py		randomlayer.py
residuallayer.py		residuallayer.py
resnet.py		resnet.py
sdr.py		sdr.py
sparsify.py		sparsify.py
spatialcrossmaplrn.py		spatialcrossmaplrn.py
stats.py		stats.py
timer.py		timer.py
vae.py		vae.py
weightnormalizedlinear.py		weightnormalizedlinear.py
xavier.py		xavier.py
zero.py		zero.py
zeromeanrrelu.py		zeromeanrrelu.py

Provide feedback