Training: Use e-lab's torch-toolbox sanitize.

This commit is contained in:
Brandon Amos 2016-01-09 17:17:20 -05:00
parent 6029baf48e
commit 8155840989
2 changed files with 96 additions and 16 deletions

95
training/sanitize.lua Normal file
View File

@ -0,0 +1,95 @@
-- From https://github.com/e-lab/torch-toolbox/blob/master/Sanitize/sanitize.lua
require('torch')
require('nn')
require('cunn')
require('cudnn')
-- common obj name to be freed
local common = {'output', 'gradInput'}
-- temporary buffer name other than output/gradInput
local t = {
-- convolution
['nn.SpatialConvolution'] = {'finput', 'fgradInput'},
['nn.SpatialConvolutionMM'] = {'finput', 'fgradInput'},
-- pooling
['nn.SpatialMaxPooling'] = {'indices'},
['nn.TemporalMaxPooling'] = {'indices'},
['nn.VolumetricMaxPooling'] = {'indices'},
['nn.SpatialFractionalMaxPooling'] = {'indices'},
-- regularizer
['nn.BatchNormalization'] = {'buffer', 'buffer2', 'centered', 'normalized'},
['nn.SpatialBatchNormalization'] = {'buffer', 'buffer2','centered', 'normalized'},
['nn.Dropout'] = {'noise'},
['nn.SpatialDropout'] = {'noise'},
-- transfer
['nn.PReLU'] = {'gradWeightBuf', 'gradWeightBuf2'},
['nn.LogSigmoid'] = {'buffer'},
-- etc
['nn.Mean'] = {'_gradInput'},
['nn.Normalize'] = {'_output', 'norm', 'normp'},
['nn.PairwiseDistance'] = {'diff'},
['nn.Reshape'] = {'_input', '_gradOutput'},
-- fbcunn
['nn.AbstractParallel'] = {'homeGradBuffers', 'input_gpu', 'gradOutput_gpu', 'gradInput_gpu'},
['nn.DataParallel'] = {'homeGradBuffers', 'input_gpu', 'gradOutput_gpu', 'gradInput_gpu'},
['nn.ModelParallel'] = {'homeGradBuffers', 'input_gpu', 'gradOutput_gpu', 'gradInput_gpu'},
}
local function free_table_or_tensor(val, name, field)
if type(val[name]) == 'table' then
val[name] = {}
elseif type(val[name]) == 'userdata' then
val[name] = field.new()
end
end
local function is_member(name, t)
if t == nil then
return false
end
for _, value in pairs(t) do
if name == value then
return true
end
end
return false
end
-- Taken and modified from Soumith's imagenet-multiGPU.torch code
-- https://github.com/soumith/imagenet-multiGPU.torch/blob/master/train.lua
local function sanitize(model)
local list = model:listModules()
for _,val in ipairs(list) do
for name,field in pairs(val) do
-- remove ffi obj
if torch.type(field) == 'cdata' then
val[name] = nil
-- remove common obj
elseif is_member(name, common) then
free_table_or_tensor(val, name, field)
-- remove specific obj
elseif is_member(name, t[val.__typename]) then
free_table_or_tensor(val, name, field)
end
end
end
return model
end
return sanitize

View File

@ -23,6 +23,7 @@ require 'torchx' --for concetration the table of tensors
paths.dofile("OpenFaceOptim.lua") paths.dofile("OpenFaceOptim.lua")
local sanitize = paths.dofile('sanitize.lua')
local optimMethod = optim.adadelta local optimMethod = optim.adadelta
local optimState = {} -- Use for other algorithms like SGD local optimState = {} -- Use for other algorithms like SGD
@ -34,22 +35,6 @@ local batchNumber
local triplet_loss local triplet_loss
local function sanitize(net)
net:apply(function (val)
for name,field in pairs(val) do
if torch.type(field) == 'cdata' then val[name] = nil end
if name == 'homeGradBuffers' then val[name] = nil end
if name == 'input_gpu' then val['input_gpu'] = {} end
if name == 'gradOutput_gpu' then val['gradOutput_gpu'] = {} end
if name == 'gradInput_gpu' then val['gradInput_gpu'] = {} end
if (name == 'output' or name == 'gradInput')
and torch.type(field) == 'torch.CudaTensor' then
cutorch.withDevice(field:getDevice(), function() val[name] = field.new() end)
end
end
end)
end
-- From https://groups.google.com/d/msg/torch7/i8sJYlgQPeA/wiHlPSa5-HYJ -- From https://groups.google.com/d/msg/torch7/i8sJYlgQPeA/wiHlPSa5-HYJ
local function replaceModules(net, orig_class_name, replacer) local function replaceModules(net, orig_class_name, replacer)
local nodes, container_nodes = net:findModules(orig_class_name) local nodes, container_nodes = net:findModules(orig_class_name)