1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
|
----------------------------------------------------------------------------------------
-- Learns a network using a single set of parameters. Typically this run against the
-- full set of training data after reasonable parameters have been found using the
-- grid search.
--
-- Authors:
-- Peter Abeles
--
----------------------------------------------------------------------
require 'pl'
require 'trepl'
require 'torch' -- torch
require 'image' -- to visualize the dataset
require 'nn' -- provides all sorts of trainable modules/layers
require 'posix.sys.stat' -- luarocks install luaposix
----------------------------------------------------------------------
print(sys.COLORS.red .. '==> processing options')
opt = lapp[[
-r,--learningRate (default 1e-3) learning rate
--sgdLearningRateDecay (default 1e-7) learning rate decay (in # samples)
--sgdWeightDecay (default 1e-5) L2 penalty on the weights
--sgdMomentum (default 0.1) momentum
--adamBeta1 (default 0.9) adam beta1 parameter
--adamBeta2 (default 0.999) adam beta2 parameter
-l,--model (default version001) which model to load
-b,--batchSize (default 128) batch size
-t,--threads (default 1) number of threads
-p,--type (default float) float or cuda
-i,--devid (default 1) device ID (if using CUDA)
-s,--size (default small) dataset: small or full or extra
-o,--save (default results) save directory
--maxNoImprovement (default 0) How long it will run since it last improved. 0 to distable
--patches (default all) percentage of samples to use for testing'
--visualize (default false) visualize dataset
--search (default sgd) optimization algorithm (sgd,adam)
]]
-- nb of threads and fixed seed (for repeatable experiments)
torch.setnumthreads(opt.threads)
torch.manualSeed(1)
torch.setdefaulttensortype('torch.FloatTensor')
print("model " .. opt.model)
print("A random number " .. torch.random(10000))
print("data type " .. opt.type)
print("search " .. opt.search)
-- type:
if opt.type == 'cuda' then
print(sys.COLORS.red .. '==> switching to CUDA')
require 'cunn'
cutorch.setDevice(opt.devid)
print(sys.COLORS.red .. '==> using GPU #' .. cutorch.getDevice())
end
package.path = package.path .. ";models/" .. opt.model .. '.lua'
----------------------------------------------------------------------
print(sys.COLORS.red .. '==> load modules')
local data = require 'data'
local ttrain = require 'train'
local ttest = require 'test'
local train = ttrain.train
local reset_train = ttrain.reset
local test = ttest.test
local reset_test = ttest.reset
----------------------------------------------------------------------
print(sys.COLORS.red .. '==> training!')
reset_train()
reset_test()
local best_test = 0
local epoc = 0
local last_best = 0
-- save all command line arguments to disk
torch.save(paths.concat(opt.save, 'arguments.t7'), opt, 'ascii')
while true do
local results_train,model_train = train(data.trainData)
local results_test = test(data.testData)
-- If the model is better then save it
if best_test < results_test then
best_test = results_test
last_best = epoc
print("Saving model. score = "..best_test)
local model_file_name = paths.concat(opt.save, 'model.net')
local model1 = model_train:clone()
torch.save(model_file_name, model1:clearState())
end
-- If performance hasn't improved in a long time stop
if opt.maxNoImprovement > 0 and epoc - last_best >= opt.maxNoImprovement then
break
else
print("Epocs since improvement "..(epoc - last_best).." max allowed "..opt.maxNoImprovement)
end
epoc = epoc + 1
end
print("Finished!")
|
