Torch7学习笔记[2] ---神经网络的主体框架

参考资料：

https://github.com/soumith/cvpr2015/blob/master/Deep%20Learning%20with%20Torch.ipynb

将整个框架分为以下几个模块：

1、options设置

2、train、test预处理以及读取

3、net结构以及criterion建立

4、train设置

5、test设置

6、保存中间结果以及断点开始(待完善）

y以上每个功能模块单独由一个文件完成，整个结构分为7个文件

main.lua opt.lua dataloder.lua model.lua train.lua test.lua checkpont.lua（待完善）

require 'torch'
require 'nn'
require 'optim'

local DataLoder = require 'dataloder'  --load the dataloder.lua
local opts = require 'opt'
local Model = require 'model'
local Test = require 'test'
local checkpoints = require 'checkpoint'
local Trainer = require 'train'

torch.setdefaulttensortype = ('torch.FloatTensor')  --
torch.setnumthreads()
torch.manualSeed(opt.manualSeed)
cutorch.manualSeedAll(opt.manualSeed)

local opt = opts.parse(arg)  --load the options
local trainset,testset = DataLoder.creat(opt) --load the dataset

local model,criterion = Model.setup(opt) --load the model,criterion
if(opt.type == 'cuda')   then  --turn on gpu:model-criterion-data-label
    model = model:cuda() 
    criterion = criterion:cuda()
    trainset.data = trainset.data:cuda()
    trainset.label = trainset.label:cuda()
    testset.data = testset.data:cuda()
    testset.label = testset.label:cuda()
end

function trainset:size() --prepare for training 
    return self.data:size() 
end
local trainer = Trainer(model,criterion,opt)

bestModel = false
for epoch = ,opt.max_epoch do
    local current_error = trainer:train(epoch,trainset)
    --save the current station
    --checkpoints.save(epoch, model, trainer.optimState, bestModel, opt)
end

local correct_rate = Test.run(opt,testset,model)
print(correct_rate)
           

运行程序时，直接在文件所在目录终端执行：th main.lua 即可运行程序。若需改变options，例如gpu运行：th main.lua –type cuda