http://cilvr.cs.nyu.edu/lib/exe/fetch.php?media=deeplearning:dl-backprop1.pdf
In [2]:
require 'nn'
require 'optim'
path = require 'pl.path'
whetlab = require('whetlab')
In [3]:
-- Classes to predict:
classes = {'0','1','2','3','4','5','6','7','8','9'}
In [4]:
iopt = {model="cnn", optimization="sgd", localData=true, cuda=false}
In [5]:
batchChoices = {50,100,250,500,1000}
whetlabJob = {momentum=0.25, batchChoice=2, dropout=0.0, layerSize=500}
whetlabJob.learningRates = {
1e-1,
1e-1,
1e-1,
1e-1,
1e-1,
1e-1,
1e-1,
1e-1
}
whetlabJob.weightDecays = {
1e-5,
1e-5,
1e-5,
1e-5,
1e-5,
1e-5,
1e-5,
1e-5
}
whetlabJob.initialScale = {
1.5e-1,
1.5e-1,
1.5e-1,
1.5e-1,
1.5e-1,
1.5e-1,
1.5e-1,
1.5e-1
}
In [6]:
opt = {
learningRate = whetlabJob.learningRate,
momentum = whetlabJob.momentum,
weightDecay = whetlabJob.weightDecay,
network = '',
cuda = iopt.cuda,
batchSize = batchChoices[whetlabJob.batchChoice],
nEpochs = 50,
loss = 'nll',
model = iopt.model,
optimization = iopt.optimization,
cnn = {
clayers = {
{1,32,5,5},
{32,64,5,5},
},
players = {
{3,3,3,3},
{2,2,2,2},
},
lunit = nn.Linear,
punit = nn.SpatialMaxPooling,
nunit = nn.ReLU,
cunit = nn.SpatialConvolutionMM,
},
linear = {dropout=whetlabJob.dropout}
}
In [7]:
startingSize = 64*2*2
opt.cnn.llayers = {
{64*2*2, whetlabJob.layerSize},
{whetlabJob.layerSize, #classes},
}
opt.cnn.dropout = whetlabJob.dropout
In [8]:
-- Do everything with floats:
torch.setdefaulttensortype('torch.FloatTensor')
-- Fix seed
torch.manualSeed(1)
-- CUDA
if opt.cuda then
require 'cunn'
cutorch.manualSeed(1)
end
-- Define model to train
model = nn.Sequential()
-- define model to train
model = nn.Sequential()
-- build cnn:
for i,clayer in ipairs(opt.cnn.clayers) do
model:add(opt.cnn.cunit(unpack(clayer)))
model:add(opt.cnn.nunit())
local player = opt.cnn.players[i]
if player then
model:add(opt.cnn.punit(unpack(player)))
end
end
model:add(nn.Reshape(opt.cnn.llayers[1][1]))
for i,llayer in ipairs(opt.cnn.llayers) do
model:add(opt.cnn.lunit(unpack(llayer)))
if i < #opt.cnn.llayers then
if opt.cnn.dropout > 0 then
model:add(nn.Dropout(opt.cnn.dropout))
end
model:add(opt.cnn.nunit())
end
end
In [9]:
-- Loss function (negative log-likelihood)
model:add(nn.LogSoftMax())
loss = nn.ClassNLLCriterion()
In [10]:
-- move model to CUDA?
if opt.cuda then
model:cuda()
loss:cuda()
end
In [11]:
-- retrieve parameters and gradients
parameters,gradParameters = model:getParameters()
In [12]:
-- Set per-layer learning rates (separate or biases and weights)
params,_ = model:parameters()
paramSizes = {0}
for i=1,#params do
paramSizes[i+1] = torch.numel(params[i])
end
boundaries = torch.cumsum(torch.Tensor(paramSizes))
start = boundaries[{{1,torch.numel(boundaries)-1}}]+1
stop = boundaries[{{2,torch.numel(boundaries)}}]
learningRates = torch.zeros(parameters:size(1))
weightDecays = torch.zeros(parameters:size(1))
for i=1,#whetlabJob.learningRates do
learningRates[{{start[i],stop[i]}}] = whetlabJob.learningRates[i]
weightDecays[{{start[i],stop[i]}}] = whetlabJob.weightDecays[i]
end
In [13]:
-- verbose
print('using model:')
print(model)
Out[13]:
using model:
nn.Sequential {
[input -> (1) -> (2) -> (3) -> (4) -> (5) -> (6) -> (7) -> (8) -> (9) -> (10) -> (11) -> output]
(1): nn.SpatialConvolutionMM
(2): nn.ReLU
(3): nn.SpatialMaxPooling
(4): nn.SpatialConvolutionMM
(5): nn.ReLU
(6): nn.SpatialMaxPooling
(7): nn.Reshape(256)
(8): nn.Linear(256 -> 500)
(9): nn.ReLU
(10): nn.Linear(500 -> 10)
(11): nn.LogSoftMax
}
{
output : FloatTensor - empty
gradInput : FloatTensor - empty
modules :
{
1 :
nn.SpatialConvolutionMM
{
padding : 0
kW : 5
nInputPlane : 1
gradBias : FloatTensor - size: 32
dW : 1
gradWeight : FloatTensor - size: 32x25
output : FloatTensor - empty
fgradInput : FloatTensor - empty
finput : FloatTensor - empty
bias : FloatTensor - size: 32
weight : FloatTensor - size: 32x25
nOutputPlane : 32
gradInput : FloatTensor - empty
kH : 5
dH : 1
}
2 :
nn.ReLU
{
val : 0
output : FloatTensor - empty
gradInput : FloatTensor - empty
threshold : 0
}
3 :
nn.SpatialMaxPooling
{
kW : 3
kH : 3
indices : FloatTensor - empty
dW : 3
gradInput : FloatTensor - empty
output : FloatTensor - empty
dH : 3
}
4 :
nn.SpatialConvolutionMM
{
padding : 0
kW : 5
nInputPlane : 32
gradBias : FloatTensor - size: 64
dW : 1
gradWeight : FloatTensor - size: 64x800
output : FloatTensor - empty
fgradInput : FloatTensor - empty
finput : FloatTensor - empty
bias : FloatTensor - size: 64
weight : FloatTensor - size: 64x800
nOutputPlane : 64
gradInput : FloatTensor - empty
kH : 5
dH : 1
}
5 :
nn.ReLU
{
val : 0
output : FloatTensor - empty
gradInput : FloatTensor - empty
threshold : 0
}
6 :
nn.SpatialMaxPooling
{
kW : 2
kH : 2
indices : FloatTensor - empty
dW : 2
gradInput : FloatTensor - empty
output : FloatTensor - empty
dH : 2
}
7 :
nn.Reshape(256)
{
_gradOutput : FloatTensor - empty
size : LongStorage - size: 1
nelement : 256
_input : FloatTensor - empty
batchsize : LongStorage - size: 2
output : FloatTensor - empty
gradInput : FloatTensor - empty
}
8 :
nn.Linear(256 -> 500)
{
bias : FloatTensor - size: 500
weight : FloatTensor - size: 500x256
gradInput : FloatTensor - empty
Out[13]:
gradBias : FloatTensor - size: 500
gradWeight : FloatTensor - size: 500x256
output : FloatTensor - empty
}
9 :
nn.ReLU
{
val : 0
output : FloatTensor - empty
gradInput : FloatTensor - empty
threshold : 0
}
10 :
nn.Linear(500 -> 10)
{
bias : FloatTensor - size: 10
weight : FloatTensor - size: 10x500
gradInput : FloatTensor - empty
gradBias : FloatTensor - size: 10
gradWeight : FloatTensor - size: 10x500
output : FloatTensor - empty
}
11 :
nn.LogSoftMax
{
gradInput : FloatTensor - empty
output : FloatTensor - empty
}
}
}
In [14]:
-- Get/create dataset
if not path.exists('mnist') then
os.execute[[
curl https://s3.amazonaws.com/torch.data/mnist.tgz -o mnist.tgz
tar xvf mnist.tgz
rm mnist.tgz
]]
end
validData = {}
trainData = torch.load('mnist/train.t7')
testData = torch.load('mnist/test.t7')
-- Update data set sizes
-- validData = testData
validData.x = trainData.x[{{50001,-1},{}}]
validData.y = trainData.y[{{50001,-1}}]
validData.size = 10000
trainData.x = trainData.x[{{1,50000},{}}]
trainData.y = trainData.y[{{1,50000}}]
trainData.size = 50000
In [15]:
-- CUDA?
if opt.cuda then
trainData.x = trainData.x:cuda()
trainData.y = trainData.y:cuda()
validData.x = validData.x:cuda()
validData.y = validData.y:cuda()
testData.x = testData.x:cuda()
testData.y = testData.y:cuda()
end
In [16]:
-- Normalize data:
-- (Note: we usually make data std=1, but with ReLU activations, much higher variance
-- accelerates learning by quite a bit)
local std = trainData.x:std()
print('normalizing (standard deviation = ' .. std .. ')')
trainData.x:mul(1/std)
validData.x:mul(1/std)
Out[16]:
normalizing (standard deviation = 0.2755747643016)
In [17]:
-- this matrix records the current confusion across classes
confusion = optim.ConfusionMatrix(classes)
In [18]:
-- a function to make a mini batch
inputs, targets = {}, {}
function makeBatch(dataset,didx)
inputs = dataset.x[{ {didx,didx+opt.batchSize-1} }]
targets = dataset.y[{ {didx,didx+opt.batchSize-1} }]
end
In [19]:
-- create closure to evaluate f(X) and df/dX
fgeval = function()
-- reset gradients
gradParameters:zero()
-- evaluate function for complete mini batch
local outputs = model:forward(inputs)
local f = loss:forward(outputs, targets)
-- estimate df/dW
local df_do = loss:backward(outputs, targets)
model:backward(inputs, df_do)
-- update confusion
for i = 1,opt.batchSize do
confusion:add(outputs[i], targets[i])
end
-- return f and df/dX
return f,gradParameters
end
In [20]:
-- create closure to evaluate f(X)
feval = function()
-- evaluate function for complete mini batch
local outputs = model:forward(inputs)
local f = loss:forward(outputs, targets)
-- update confusion
for i = 1,opt.batchSize do
confusion:add(outputs[i], targets[i])
end
-- return f and df/dX
return f
end
In [21]:
-- Find out the indices of layers with weights
acc = {}
for i=1,#model.modules do
if model.modules[i].weight then
acc[#acc+1] = i
end
end
-- Now initialize those layers in a dope manner
for i=1,#acc do
layerIdx = acc[i]
model.modules[layerIdx]:reset()
weightScale = whetlabJob.initialScale[(i*2)-1]
biasScale = whetlabJob.initialScale[(i*2)]
model.modules[layerIdx].weight:mul(weightScale)
model.modules[layerIdx].bias:mul(biasScale)
end
In [22]:
-- Use simple SGD (first-order method):
optimState = {
learningRate = 1.0, -- learningRates scales the learningRate, which defaults to 1e-3.
learningRates = learningRates,
momentum = opt.momentum,
weightDecays = weightDecays,
}
optimFunc = optim.sgd
In [23]:
-- training function
function train(epoch)
-- restore dropout
model:training()
-- do one epoch
print('')
print('on training set:')
print("online epoch # " .. epoch .. ' [batchSize = ' .. opt.batchSize .. ']')
-- loop:
for t = 1,trainData.size,opt.batchSize do
-- Data idx:
makeBatch(trainData, t)
-- Perform SGD step:
optimFunc(fgeval, parameters, optimState)
-- disp progress
-- xlua.progress(t, trainData.size)
end
-- print confusion matrix
print('')
print(confusion)
confusion:zero()
end
In [24]:
-- test function
perf = 0
noProgress = 0
backwardProgress = 0
function test(epoch)
-- shutdown dropout for testing
model:evaluate()
-- test over given dataset
print('on testing Set:')
for t = 1,validData.size,opt.batchSize do
-- create mini batch
makeBatch(validData, t)
-- evaluate f(x)
feval()
end
-- print confusion matrix
print('')
print(confusion)
-- track performance and adjust learning rate (sgd only:
if opt.optimization == 'sgd' then
if confusion.totalValid < (perf + .001) then
optimState.learningRates = torch.clamp(optimState.learningRates/2.0, 1e-5, math.huge)
end
if confusion.totalValid < perf and optimState.learningRate == 1e-5 then
backwardProgress = backwardProgress + 1
elseif confusion.totalValid < (perf + .001) and optimState.learningRate == 1e-5 then
noProgress = noProgress + 1
else
noProgress = 0
backwardProgress = 0
end
end
perf = confusion.totalValid
-- log results, and model:
-- logScore(epoch)
print("\n\n")
print("No Progress Iter: " .. noProgress)
print("Backward Progress Iter: " .. backwardProgress)
print("\n\n")
-- reset
confusion:zero()
return perf
end
In [25]:
-- and train!
for epoch = 1,40 do -- opt.nEpochs do
-- train:
train(epoch)
-- test:
perf = test(epoch)
end
Out[25]:
on training set:
online epoch # 1 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 180 4751 1 0 0 0 0 0 0 0] 3.650% [class: 0]
[ 5 5672 1 0 0 0 0 0 0 0] 99.894% [class: 1]
[ 106 4851 11 0 0 0 0 0 0 0] 0.221% [class: 2]
[ 88 4996 17 0 0 0 0 0 0 0] 0.000% [class: 3]
[ 108 4747 4 0 0 0 0 0 0 0] 0.000% [class: 4]
[ 89 4411 6 0 0 0 0 0 0 0] 0.000% [class: 5]
[ 107 4841 3 0 0 0 0 0 0 0] 0.000% [class: 6]
[ 62 5092 12 0 0 0 0 9 0 0] 0.174% [class: 7]
[ 77 4748 15 2 0 0 0 0 0 0] 0.000% [class: 8]
[ 101 4881 5 1 0 0 0 0 0 0]] 0.000% [class: 9]
+ average row correct: 10.393929191632%
+ average rowUcol correct (VOC measure): 1.5140775113832%
+ global correct: 11.744%
{
averageUnionValid : 0.015140775113832
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.11744
_prediction : FloatTensor - size: 10
averageValid : 0.10393929191632
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 988 3 0 0 0 0 0 0 0 0] 99.697% [class: 0]
[ 37 1011 9 7 0 0 0 0 0 0] 95.019% [class: 1]
[ 933 25 28 4 0 0 0 0 0 0] 2.828% [class: 2]
[ 944 47 4 35 0 0 0 0 0 0] 3.398% [class: 3]
[ 936 40 0 7 0 0 0 0 0 0] 0.000% [class: 4]
[ 853 58 4 0 0 0 0 0 0 0] 0.000% [class: 5]
[ 899 55 13 0 0 0 0 0 0 0] 0.000% [class: 6]
[ 906 119 5 56 0 0 0 4 0 0] 0.367% [class: 7]
[ 869 103 23 14 0 0 0 0 0 0] 0.000% [class: 8]
[ 873 67 0 21 0 0 0 0 0 0]] 0.000% [class: 9]
+ average row correct: 20.130938561633%
+ average rowUcol correct (VOC measure): 8.2047302555293%
+ global correct: 20.66%
{
averageUnionValid : 0.082047302555293
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
Out[25]:
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.2066
_prediction : FloatTensor - size: 10
averageValid : 0.20130938561633
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 2 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4529 19 20 31 29 62 60 111 60 11] 91.829% [class: 0]
[ 5 5464 41 60 12 7 18 26 41 4] 96.231% [class: 1]
[ 290 191 3906 161 72 14 103 115 103 13] 78.623% [class: 2]
[ 236 185 254 3747 31 151 39 241 160 57] 73.456% [class: 3]
[ 303 35 26 9 3835 11 153 122 38 327] 78.926% [class: 4]
[ 343 129 57 186 90 3080 182 105 229 105] 68.353% [class: 5]
[ 305 163 44 14 148 83 4102 31 49 12] 82.852% [class: 6]
[ 209 118 88 93 72 10 5 4241 41 298] 81.952% [class: 7]
[ 263 262 106 227 91 210 112 144 3289 138] 67.926% [class: 8]
[ 245 82 23 66 410 69 20 408 81 3584]] 71.852% [class: 9]
+ average row correct: 79.200088381767%
+ average rowUcol correct (VOC measure): 65.827092528343%
+ global correct: 79.554%
{
averageUnionValid : 0.65827092528343
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.79554
_prediction : FloatTensor - size: 10
averageValid : 0.79200088381767
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 978 0 1 0 0 0 4 4 0 4] 98.688% [class: 0]
[ 0 1060 1 0 0 0 0 3 0 0] 99.624% [class: 1]
[ 5 11 902 9 2 0 1 54 5 1] 91.111% [class: 2]
[ 1 0 2 1002 1 7 0 7 1 9] 97.282% [class: 3]
[ 1 8 0 0 942 0 9 2 0 21] 95.829% [class: 4]
[ 11 0 0 17 3 853 17 3 3 8] 93.224% [class: 5]
[ 5 2 0 0 4 1 954 0 1 0] 98.656% [class: 6]
[ 1 4 0 1 1 0 0 1079 0 4] 98.991% [class: 7]
[ 12 15 3 16 4 3 11 13 905 27] 89.693% [class: 8]
[ 4 3 0 3 12 2 0 32 0 905]] 94.173% [class: 9]
+ average row correct: 95.727001428604%
+ average rowUcol correct (VOC measure): 91.951560378075%
+ global correct: 95.8%
{
averageUnionValid : 0.91951560378075
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
Out[25]:
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.958
_prediction : FloatTensor - size: 10
averageValid : 0.95727001428604
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 3 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4844 3 10 1 3 12 32 1 17 9] 98.216% [class: 0]
[ 1 5589 29 8 6 0 4 13 24 4] 98.433% [class: 1]
[ 10 18 4786 44 12 3 2 59 32 2] 96.337% [class: 2]
[ 4 6 47 4911 1 39 3 29 36 25] 96.275% [class: 3]
[ 6 4 8 0 4717 0 31 15 9 69] 97.078% [class: 4]
[ 13 2 3 36 3 4377 24 5 29 14] 97.137% [class: 5]
[ 28 6 2 2 19 22 4852 0 20 0] 98.000% [class: 6]
[ 6 15 61 13 12 5 0 5003 7 53] 96.676% [class: 7]
[ 13 27 28 35 19 27 20 10 4613 50] 95.271% [class: 8]
[ 27 5 3 28 49 19 3 54 29 4771]] 95.650% [class: 9]
+ average row correct: 96.907165646553%
+ average rowUcol correct (VOC measure): 94.01674926281%
+ global correct: 96.926%
{
averageUnionValid : 0.9401674926281
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
Out[25]:
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.96926
_prediction : FloatTensor - size: 10
averageValid : 0.96907165646553
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 981 0 2 0 0 0 3 1 1 3] 98.991% [class: 0]
[ 0 1060 1 0 0 0 1 2 0 0] 99.624% [class: 1]
[ 1 8 954 0 3 0 1 21 1 1] 96.364% [class: 2]
[ 2 1 5 988 1 11 0 5 6 11] 95.922% [class: 3]
[ 0 6 0 0 958 0 9 2 1 7] 97.457% [class: 4]
[ 4 0 0 3 1 891 14 1 0 1] 97.377% [class: 5]
[ 1 0 0 0 0 1 965 0 0 0] 99.793% [class: 6]
[ 0 4 0 0 1 0 0 1078 2 5] 98.899% [class: 7]
[ 4 3 1 3 1 5 17 4 963 8] 95.441% [class: 8]
[ 4 3 0 3 13 5 0 12 2 919]] 95.630% [class: 9]
+ average row correct: 97.549760341644%
+ average rowUcol correct (VOC measure): 95.243304371834%
+ global correct: 97.57%
{
averageUnionValid : 0.95243304371834
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9757
_prediction : FloatTensor - size: 10
averageValid : 0.97549760341644
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 4 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4878 3 8 1 5 3 17 0 10 7] 98.905% [class: 0]
[ 1 5626 16 4 2 0 3 12 11 3] 99.084% [class: 1]
[ 11 10 4856 25 5 1 2 39 18 1] 97.746% [class: 2]
[ 3 2 29 4985 0 27 1 17 18 19] 97.726% [class: 3]
[ 7 2 5 0 4779 1 15 6 8 36] 98.354% [class: 4]
[ 10 2 3 20 3 4419 16 3 23 7] 98.069% [class: 5]
[ 20 5 3 1 8 12 4889 0 13 0] 98.748% [class: 6]
[ 3 15 35 10 11 2 0 5060 7 32] 97.778% [class: 7]
[ 10 21 22 22 11 14 17 9 4686 30] 96.778% [class: 8]
[ 18 1 1 15 31 12 3 39 19 4849]] 97.213% [class: 9]
+ average row correct: 98.040062189102%
+ average rowUcol correct (VOC measure): 96.167304515839%
+ global correct: 98.054%
{
averageUnionValid : 0.96167304515839
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.98054
_prediction : FloatTensor - size: 10
averageValid : 0.98040062189102
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 981 0 1 0 0 1 3 1 1 3] 98.991% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 6 970 0 0 0 1 11 1 0] 97.980% [class: 2]
[ 1 0 4 1004 1 12 0 2 4 2] 97.476% [class: 3]
[ 0 5 0 0 959 1 8 3 1 6] 97.558% [class: 4]
[ 4 0 0 3 1 893 13 0 0 1] 97.596% [class: 5]
[ 0 0 0 0 0 2 965 0 0 0] 99.793% [class: 6]
[ 0 3 1 0 0 0 0 1083 2 1] 99.358% [class: 7]
[ 3 3 2 3 1 7 20 0 966 4] 95.738% [class: 8]
[ 3 3 0 2 10 10 0 8 1 924]] 96.150% [class: 9]
+ average row correct: 98.026378750801%
+ average rowUcol correct (VOC measure): 96.138559579849%
+ global correct: 98.05%
{
averageUnionValid : 0.96138559579849
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9805
_prediction : FloatTensor - size: 10
averageValid : 0.98026378750801
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 5 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4892 3 4 1 3 4 13 2 4 6] 99.189% [class: 0]
[ 1 5642 10 3 2 0 1 10 7 2] 99.366% [class: 1]
[ 5 8 4887 15 5 0 2 31 14 1] 98.370% [class: 2]
[ 2 1 22 5013 0 22 1 15 12 13] 98.275% [class: 3]
[ 4 1 5 0 4798 2 11 6 8 24] 98.745% [class: 4]
[ 5 2 4 20 1 4434 12 4 18 6] 98.402% [class: 5]
[ 18 4 1 0 7 9 4898 0 14 0] 98.930% [class: 6]
[ 3 12 28 8 10 1 0 5087 7 19] 98.300% [class: 7]
[ 7 16 19 18 10 13 14 5 4718 22] 97.439% [class: 8]
[ 10 1 0 11 24 10 2 30 15 4885]] 97.935% [class: 9]
+ average row correct: 98.49492251873%
+ average rowUcol correct (VOC measure): 97.044394016266%
+ global correct: 98.508%
{
averageUnionValid : 0.97044394016266
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.98508
_prediction : FloatTensor - size: 10
averageValid : 0.9849492251873
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 980 0 1 0 0 2 3 1 1 3] 98.890% [class: 0]
[ 0 1061 0 0 0 0 2 1 0 0] 99.718% [class: 1]
[ 1 4 976 0 1 0 1 7 0 0] 98.586% [class: 2]
[ 1 0 5 1006 0 13 0 1 3 1] 97.670% [class: 3]
[ 0 7 0 0 960 1 4 3 1 7] 97.660% [class: 4]
[ 2 0 1 3 0 895 13 0 0 1] 97.814% [class: 5]
[ 0 0 0 0 0 1 966 0 0 0] 99.897% [class: 6]
[ 0 3 1 0 0 0 0 1082 2 2] 99.266% [class: 7]
[ 3 3 3 2 1 6 15 0 971 5] 96.234% [class: 8]
[ 3 3 0 2 10 6 0 6 1 930]] 96.774% [class: 9]
+ average row correct: 98.250897526741%
+ average rowUcol correct (VOC measure): 96.564480662346%
+ global correct: 98.27%
{
averageUnionValid : 0.96564480662346
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
Out[25]:
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9827
_prediction : FloatTensor - size: 10
averageValid : 0.98250897526741
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 6 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4908 2 3 0 2 2 8 1 1 5] 99.513% [class: 0]
[ 0 5645 5 3 3 0 2 10 8 2] 99.419% [class: 1]
[ 5 6 4912 8 5 0 2 20 9 1] 98.873% [class: 2]
[ 2 1 18 5030 0 17 1 11 8 13] 98.608% [class: 3]
[ 2 1 5 0 4813 1 10 4 6 17] 99.053% [class: 4]
[ 2 1 3 16 1 4455 12 2 11 3] 98.868% [class: 5]
[ 15 3 1 0 6 8 4906 0 12 0] 99.091% [class: 6]
[ 3 12 21 6 10 0 0 5103 6 14] 98.609% [class: 7]
[ 5 10 13 14 4 10 14 5 4752 15] 98.141% [class: 8]
[ 8 2 0 7 19 11 1 23 13 4904]] 98.316% [class: 9]
+ average row correct: 98.849158883095%
+ average rowUcol correct (VOC measure): 97.727466821671%
+ global correct: 98.856%
{
averageUnionValid : 0.97727466821671
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.98856
_prediction : FloatTensor - size: 10
averageValid : 0.98849158883095
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 975 0 2 0 0 5 4 1 1 3] 98.385% [class: 0]
[ 0 1061 0 0 0 0 2 1 0 0] 99.718% [class: 1]
[ 1 4 977 0 1 0 1 6 0 0] 98.687% [class: 2]
[ 0 0 5 1008 0 12 0 1 4 0] 97.864% [class: 3]
[ 0 7 0 0 961 1 3 3 1 7] 97.762% [class: 4]
[ 1 0 0 3 0 897 12 0 1 1] 98.033% [class: 5]
[ 0 0 0 0 0 1 966 0 0 0] 99.897% [class: 6]
[ 0 3 1 0 0 0 0 1083 1 2] 99.358% [class: 7]
[ 3 3 3 2 0 6 12 0 976 4] 96.729% [class: 8]
[ 3 1 0 1 10 6 0 3 1 936]] 97.399% [class: 9]
+ average row correct: 98.383156061172%
+ average rowUcol correct (VOC measure): 96.814914941788%
+ global correct: 98.4%
{
averageUnionValid : 0.96814914941788
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.984
_prediction : FloatTensor - size: 10
averageValid : 0.98383156061172
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 7 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4911 1 2 0 2 0 8 0 3 5] 99.574% [class: 0]
[ 0 5652 3 1 3 0 1 10 7 1] 99.542% [class: 1]
[ 4 5 4922 8 4 0 2 14 8 1] 99.074% [class: 2]
[ 2 0 17 5043 0 12 1 8 10 8] 98.863% [class: 3]
[ 1 1 5 0 4818 1 8 6 3 16] 99.156% [class: 4]
[ 2 1 2 10 1 4466 9 2 11 2] 99.112% [class: 5]
[ 12 2 2 0 5 7 4915 0 8 0] 99.273% [class: 6]
[ 3 11 17 4 7 0 0 5115 6 12] 98.841% [class: 7]
[ 3 8 9 14 2 10 8 4 4771 13] 98.534% [class: 8]
[ 6 2 0 6 18 9 1 21 10 4915]] 98.536% [class: 9]
+ average row correct: 99.050545692444%
+ average rowUcol correct (VOC measure): 98.121281266212%
+ global correct: 99.056%
{
averageUnionValid : 0.98121281266212
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.99056
_prediction : FloatTensor - size: 10
averageValid : 0.99050545692444
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 973 0 3 0 0 6 4 1 1 3] 98.184% [class: 0]
[ 0 1061 0 0 0 0 2 1 0 0] 99.718% [class: 1]
[ 1 3 978 0 1 0 1 6 0 0] 98.788% [class: 2]
[ 0 0 3 1017 0 8 0 0 2 0] 98.738% [class: 3]
[ 0 7 0 0 963 0 2 3 1 7] 97.965% [class: 4]
[ 1 0 0 3 0 899 10 0 1 1] 98.251% [class: 5]
[ 0 0 0 0 0 1 966 0 0 0] 99.897% [class: 6]
[ 0 3 1 0 0 0 0 1082 1 3] 99.266% [class: 7]
[ 3 2 1 3 0 4 11 0 981 4] 97.225% [class: 8]
[ 3 0 0 1 8 4 0 2 1 942]] 98.023% [class: 9]
+ average row correct: 98.605473041534%
+ average rowUcol correct (VOC measure): 97.247789502144%
+ global correct: 98.62%
{
averageUnionValid : 0.97247789502144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9862
_prediction : FloatTensor - size: 10
averageValid : 0.98605473041534
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 8 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4919 1 1 0 2 0 3 1 3 2] 99.736% [class: 0]
[ 0 5658 2 1 2 0 1 9 5 0] 99.648% [class: 1]
[ 2 4 4936 3 3 0 2 11 6 1] 99.356% [class: 2]
[ 2 0 11 5060 0 10 0 7 7 4] 99.196% [class: 3]
[ 1 1 4 0 4825 0 7 6 2 13] 99.300% [class: 4]
[ 3 1 1 4 1 4474 7 2 11 2] 99.290% [class: 5]
[ 10 2 3 0 4 6 4919 0 7 0] 99.354% [class: 6]
[ 3 12 15 4 7 0 0 5121 5 8] 98.957% [class: 7]
[ 2 7 6 8 4 9 5 4 4788 9] 98.885% [class: 8]
[ 3 2 0 5 16 6 1 20 10 4925]] 98.737% [class: 9]
+ average row correct: 99.245830774307%
+ average rowUcol correct (VOC measure): 98.505631089211%
+ global correct: 99.25%
{
averageUnionValid : 0.98505631089211
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
Out[25]:
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9925
_prediction : FloatTensor - size: 10
averageValid : 0.99245830774307
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 973 0 3 0 0 6 4 1 1 3] 98.184% [class: 0]
[ 0 1058 1 0 0 0 3 1 1 0] 99.436% [class: 1]
[ 1 2 980 1 1 0 1 4 0 0] 98.990% [class: 2]
[ 0 0 2 1024 0 4 0 0 0 0] 99.417% [class: 3]
[ 0 5 0 0 960 1 4 3 2 8] 97.660% [class: 4]
[ 1 0 0 3 0 900 9 0 1 1] 98.361% [class: 5]
[ 0 0 0 0 0 2 965 0 0 0] 99.793% [class: 6]
[ 0 3 0 0 0 0 0 1084 1 2] 99.450% [class: 7]
[ 2 0 1 4 0 5 9 0 984 4] 97.522% [class: 8]
[ 3 0 0 2 9 3 0 3 2 939]] 97.711% [class: 9]
+ average row correct: 98.652373552322%
+ average rowUcol correct (VOC measure): 97.341857552528%
+ global correct: 98.67%
{
averageUnionValid : 0.97341857552528
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9867
_prediction : FloatTensor - size: 10
averageValid : 0.98652373552322
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 9 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4920 0 1 0 1 0 6 1 2 1] 99.757% [class: 0]
[ 0 5664 2 0 0 0 1 9 2 0] 99.753% [class: 1]
[ 2 4 4941 3 2 1 0 9 5 1] 99.457% [class: 2]
[ 1 0 9 5068 0 7 0 6 8 2] 99.353% [class: 3]
[ 1 1 3 0 4833 0 6 2 1 12] 99.465% [class: 4]
[ 1 1 1 5 0 4484 4 1 8 1] 99.512% [class: 5]
[ 7 0 2 0 3 3 4932 0 4 0] 99.616% [class: 6]
[ 0 8 12 2 1 1 0 5139 3 9] 99.304% [class: 7]
[ 2 4 8 5 2 8 3 1 4804 5] 99.215% [class: 8]
[ 4 2 0 1 10 5 0 14 4 4948]] 99.198% [class: 9]
+ average row correct: 99.463024735451%
+ average rowUcol correct (VOC measure): 98.933103680611%
+ global correct: 99.466%
{
averageUnionValid : 0.98933103680611
_targ_idx : LongTensor - empty
Out[25]:
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.99466
_prediction : FloatTensor - size: 10
averageValid : 0.99463024735451
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 981 0 1 1 0 1 3 1 1 2] 98.991% [class: 0]
[ 0 1057 2 0 0 0 3 1 1 0] 99.342% [class: 1]
[ 1 2 983 2 0 0 1 1 0 0] 99.293% [class: 2]
[ 0 0 3 1022 0 3 0 0 2 0] 99.223% [class: 3]
[ 0 2 0 0 967 0 2 4 1 7] 98.372% [class: 4]
[ 1 0 1 3 0 900 8 0 1 1] 98.361% [class: 5]
[ 0 0 0 0 0 1 966 0 0 0] 99.897% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 2 0 1 2 0 3 7 0 992 2] 98.315% [class: 8]
[ 3 0 0 4 8 2 0 6 3 935]] 97.294% [class: 9]
+ average row correct: 98.881324529648%
+ average rowUcol correct (VOC measure): 97.800331115723%
+ global correct: 98.9%
{
averageUnionValid : 0.97800331115723
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.989
_prediction : FloatTensor - size: 10
averageValid : 0.98881324529648
}
No Progress Iter: 0
Backward Progress Iter: 0
Out[25]:
on training set:
online epoch # 10 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4925 0 0 0 1 0 3 0 2 1] 99.858% [class: 0]
[ 0 5664 1 0 0 0 0 9 4 0] 99.753% [class: 1]
[ 2 2 4948 1 2 0 2 7 4 0] 99.597% [class: 2]
[ 1 1 4 5077 0 6 0 5 5 2] 99.530% [class: 3]
[ 1 1 3 0 4839 0 5 2 1 7] 99.588% [class: 4]
[ 1 1 1 4 0 4487 4 1 6 1] 99.578% [class: 5]
[ 7 0 0 0 4 3 4934 0 3 0] 99.657% [class: 6]
[ 0 8 13 2 1 0 0 5145 1 5] 99.420% [class: 7]
[ 1 5 5 3 1 7 3 0 4815 2] 99.442% [class: 8]
[ 3 1 0 1 6 2 0 12 4 4959]] 99.419% [class: 9]
+ average row correct: 99.584307670593%
+ average rowUcol correct (VOC measure): 99.173460006714%
+ global correct: 99.586%
{
averageUnionValid : 0.99173460006714
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.99586
_prediction : FloatTensor - size: 10
averageValid : 0.99584307670593
}
Out[25]:
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 982 0 1 1 0 1 2 1 1 2] 99.092% [class: 0]
[ 0 1057 2 0 0 0 3 1 1 0] 99.342% [class: 1]
[ 1 2 983 2 0 0 1 1 0 0] 99.293% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 1 967 0 1 4 1 7] 98.372% [class: 4]
[ 1 0 1 4 0 900 8 0 1 0] 98.361% [class: 5]
[ 0 0 0 0 0 1 966 0 0 0] 99.897% [class: 6]
[ 0 2 0 1 0 0 0 1086 0 1] 99.633% [class: 7]
[ 2 0 1 2 0 3 7 1 991 2] 98.216% [class: 8]
[ 3 0 0 6 8 3 0 6 3 932]] 96.982% [class: 9]
+ average row correct: 98.850821852684%
+ average rowUcol correct (VOC measure): 97.743374109268%
+ global correct: 98.87%
{
Out[25]:
averageUnionValid : 0.97743374109268
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9887
_prediction : FloatTensor - size: 10
averageValid : 0.98850821852684
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 11 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4925 0 1 0 1 0 3 0 2 0] 99.858% [class: 0]
[ 0 5666 0 0 0 0 1 9 2 0] 99.789% [class: 1]
[ 1 3 4952 1 1 0 0 6 3 1] 99.678% [class: 2]
[ 1 1 3 5086 0 3 0 2 3 2] 99.706% [class: 3]
[ 1 1 3 0 4844 0 3 0 0 7] 99.691% [class: 4]
[ 0 1 1 5 0 4492 3 0 3 1] 99.689% [class: 5]
[ 5 0 1 0 4 3 4935 0 3 0] 99.677% [class: 6]
[ 0 7 8 0 0 0 0 5156 0 4] 99.633% [class: 7]
[ 0 3 4 2 0 6 2 0 4823 2] 99.608% [class: 8]
[ 2 1 0 0 7 1 0 9 4 4964]] 99.519% [class: 9]
+ average row correct: 99.684733748436%
+ average rowUcol correct (VOC measure): 99.373205900192%
+ global correct: 99.686%
{
averageUnionValid : 0.99373205900192
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
Out[25]:
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.99686
_prediction : FloatTensor - size: 10
averageValid : 0.99684733748436
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 1 0 1 0 1 1 2] 99.294% [class: 0]
[ 0 1057 2 0 0 0 3 1 1 0] 99.342% [class: 1]
[ 1 2 982 2 0 0 1 2 0 0] 99.192% [class: 2]
[ 0 0 2 1024 0 2 0 0 2 0] 99.417% [class: 3]
[ 0 2 0 0 968 0 1 4 1 7] 98.474% [class: 4]
[ 1 0 1 4 1 898 8 0 1 1] 98.142% [class: 5]
[ 0 0 0 0 0 1 966 0 0 0] 99.897% [class: 6]
[ 0 0 0 1 0 0 0 1089 0 0] 99.908% [class: 7]
[ 3 0 1 2 0 3 6 1 991 2] 98.216% [class: 8]
[ 3 0 0 5 8 1 0 6 3 935]] 97.294% [class: 9]
+ average row correct: 98.917667269707%
+ average rowUcol correct (VOC measure): 97.879247665405%
+ global correct: 98.94%
{
averageUnionValid : 0.97879247665405
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9894
_prediction : FloatTensor - size: 10
averageValid : 0.98917667269707
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 12 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4928 0 1 0 0 0 1 0 2 0] 99.919% [class: 0]
[ 0 5667 0 0 0 0 1 8 2 0] 99.806% [class: 1]
[ 0 2 4958 0 0 0 1 4 2 1] 99.799% [class: 2]
[ 0 1 1 5089 0 3 0 2 3 2] 99.765% [class: 3]
[ 1 1 3 0 4847 0 3 0 0 4] 99.753% [class: 4]
[ 0 1 1 4 0 4494 3 0 2 1] 99.734% [class: 5]
[ 4 0 1 0 5 2 4937 0 2 0] 99.717% [class: 6]
[ 0 7 7 0 2 0 0 5155 0 4] 99.614% [class: 7]
[ 0 3 2 2 0 4 2 0 4828 1] 99.711% [class: 8]
[ 2 1 0 0 6 2 0 8 3 4966]] 99.559% [class: 9]
+ average row correct: 99.737591147423%
+ average rowUcol correct (VOC measure): 99.477725625038%
+ global correct: 99.738%
{
averageUnionValid : 0.99477725625038
_targ_idx : LongTensor - empty
Out[25]:
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.99738
_prediction : FloatTensor - size: 10
averageValid : 0.99737591147423
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1058 1 0 0 0 3 1 1 0] 99.436% [class: 1]
[ 1 2 982 2 0 0 1 2 0 0] 99.192% [class: 2]
[ 0 0 2 1024 0 2 0 0 2 0] 99.417% [class: 3]
[ 0 2 0 0 968 0 1 4 1 7] 98.474% [class: 4]
[ 1 0 0 4 2 897 8 0 2 1] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 0 0 1 0 0 0 1089 0 0] 99.908% [class: 7]
[ 3 0 1 2 0 1 2 1 996 3] 98.712% [class: 8]
[ 3 0 0 4 7 1 0 6 3 937]] 97.503% [class: 9]
+ average row correct: 98.965820074081%
+ average rowUcol correct (VOC measure): 97.977915406227%
+ global correct: 98.99%
{
averageUnionValid : 0.97977915406227
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
Out[25]:
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9899
_prediction : FloatTensor - size: 10
averageValid : 0.98965820074081
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 13 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5667 0 0 0 0 1 8 2 0] 99.806% [class: 1]
[ 0 2 4958 0 0 0 2 4 2 0] 99.799% [class: 2]
[ 0 1 1 5090 0 2 0 2 3 2] 99.784% [class: 3]
[ 1 1 2 0 4846 0 3 2 0 4] 99.732% [class: 4]
[ 0 1 1 4 0 4496 3 0 0 1] 99.778% [class: 5]
[ 4 0 1 0 4 1 4939 0 2 0] 99.758% [class: 6]
[ 0 7 7 0 1 0 0 5158 0 2] 99.671% [class: 7]
[ 0 3 2 2 0 4 2 0 4828 1] 99.711% [class: 8]
[ 2 1 0 0 5 2 0 8 2 4968]] 99.599% [class: 9]
+ average row correct: 99.757805466652%
+ average rowUcol correct (VOC measure): 99.518758058548%
+ global correct: 99.758%
{
averageUnionValid : 0.99518758058548
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
Out[25]:
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.99758
_prediction : FloatTensor - size: 10
averageValid : 0.99757805466652
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 980 2 1 0 1 3 0 0] 98.990% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 969 0 1 3 1 7] 98.576% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 1 0 0 0 1086 0 1] 99.633% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 4 7 1 0 4 3 939]] 97.711% [class: 9]
+ average row correct: 98.948255777359%
+ average rowUcol correct (VOC measure): 97.938922047615%
+ global correct: 98.97%
{
averageUnionValid : 0.97938922047615
_targ_idx : LongTensor - empty
Out[25]:
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9897
_prediction : FloatTensor - size: 10
averageValid : 0.98948255777359
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 14 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 1 2 4958 0 0 0 2 4 1 0] 99.799% [class: 2]
[ 0 1 1 5090 0 2 0 2 3 2] 99.784% [class: 3]
[ 1 1 1 0 4847 0 3 2 0 4] 99.753% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 4 1 4940 0 2 0] 99.778% [class: 6]
[ 0 5 7 0 1 0 0 5160 0 2] 99.710% [class: 7]
[ 0 2 2 2 0 4 2 0 4829 1] 99.732% [class: 8]
[ 2 1 0 0 5 2 0 7 2 4969]] 99.619% [class: 9]
+ average row correct: 99.775559902191%
+ average rowUcol correct (VOC measure): 99.553617835045%
+ global correct: 99.776%
{
averageUnionValid : 0.99553617835045
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.99776
_prediction : FloatTensor - size: 10
averageValid : 0.99775559902191
}
Out[25]:
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 980 2 1 0 1 3 0 0] 98.990% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1086 0 2] 99.633% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 2 7 1 0 3 3 942]] 98.023% [class: 9]
+ average row correct: 98.989645838737%
+ average rowUcol correct (VOC measure): 98.016051650047%
+ global correct: 99.01%
{
averageUnionValid : 0.98016051650047
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
Out[25]:
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9901
_prediction : FloatTensor - size: 10
averageValid : 0.98989645838737
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
Out[25]:
on training set:
online epoch # 15 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5091 0 2 0 1 4 2] 99.804% [class: 3]
[ 1 0 1 0 4847 0 3 3 0 4] 99.753% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5157 0 3] 99.652% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.771655797958%
+ average rowUcol correct (VOC measure): 99.545434117317%
+ global correct: 99.772%
{
averageUnionValid : 0.99545434117317
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
Out[25]:
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.99772
_prediction : FloatTensor - size: 10
averageValid : 0.99771655797958
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 980 2 1 0 1 3 0 0] 98.990% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1086 0 2] 99.633% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 3 3 943]] 98.127% [class: 9]
+ average row correct: 99.000052213669%
+ average rowUcol correct (VOC measure): 98.035752177238%
+ global correct: 99.02%
{
averageUnionValid : 0.98035752177238
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.99000052213669
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 16 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 1 0 4848 0 3 2 0 4] 99.774% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5157 0 3] 99.652% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.775674343109%
+ average rowUcol correct (VOC measure): 99.553405642509%
+ global correct: 99.776%
{
averageUnionValid : 0.99553405642509
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
Out[25]:
totalValid : 0.99776
_prediction : FloatTensor - size: 10
averageValid : 0.99775674343109
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 980 2 1 0 1 3 0 0] 98.990% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 3 3 943]] 98.127% [class: 9]
+ average row correct: 99.009226560593%
+ average rowUcol correct (VOC measure): 98.054683208466%
+ global correct: 99.03%
{
averageUnionValid : 0.98054683208466
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99009226560593
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 17 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 1 0 4848 0 3 2 0 4] 99.774% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.777606725693%
+ average rowUcol correct (VOC measure): 99.557319283485%
+ global correct: 99.778%
{
averageUnionValid : 0.99557319283485
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
Out[25]:
totalValid : 0.99778
_prediction : FloatTensor - size: 10
averageValid : 0.99777606725693
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 980 2 1 0 1 3 0 0] 98.990% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.998820185661%
+ average rowUcol correct (VOC measure): 98.035499453545%
+ global correct: 99.02%
{
averageUnionValid : 0.98035499453545
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
Out[25]:
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.98998820185661
}
No Progress Iter: 0
Backward Progress Iter: 0
Out[25]:
on training set:
online epoch # 18 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
Out[25]:
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99008921384811
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
Out[25]:
on training set:
online epoch # 19 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
Out[25]:
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
Out[25]:
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99008921384811
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 20 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
Out[25]:
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
Out[25]:
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99008921384811
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 21 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
Out[25]:
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
Out[25]:
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99008921384811
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
Out[25]:
on training set:
online epoch # 22 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
Out[25]:
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
Out[25]:
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99008921384811
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 23 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
Out[25]:
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
Out[25]:
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99008921384811
}
No Progress Iter: 0
Backward Progress Iter: 0
Out[25]:
on training set:
online epoch # 24 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
Out[25]:
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
Out[25]:
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99008921384811
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 25 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
Out[25]:
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99008921384811
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 26 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
Out[25]:
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
Out[25]:
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99008921384811
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 27 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
Out[25]:
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9903
_prediction : FloatTensor - size: 10
Out[25]:
averageValid : 0.99008921384811
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 28 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
Out[25]:
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
Out[25]:
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
Out[25]:
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99008921384811
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 29 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1023 0 3 0 0 2 0] 99.320% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 99.008921384811%
+ average rowUcol correct (VOC measure): 98.054509162903%
+ global correct: 99.03%
{
averageUnionValid : 0.98054509162903
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
Out[25]:
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9903
_prediction : FloatTensor - size: 10
averageValid : 0.99008921384811
}
No Progress Iter: 0
Backward Progress Iter: 0
Out[25]:
on training set:
online epoch # 30 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
Out[25]:
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1022 0 3 0 0 3 0] 99.223% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.99921298027%
+ average rowUcol correct (VOC measure): 98.03530216217%
+ global correct: 99.02%
{
Out[25]:
averageUnionValid : 0.9803530216217
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.9899921298027
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 31 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
Out[25]:
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1022 0 3 0 0 3 0] 99.223% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.99921298027%
+ average rowUcol correct (VOC measure): 98.03530216217%
+ global correct: 99.02%
{
averageUnionValid : 0.9803530216217
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.9899921298027
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 32 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
Out[25]:
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1022 0 3 0 0 3 0] 99.223% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.99921298027%
+ average rowUcol correct (VOC measure): 98.03530216217%
+ global correct: 99.02%
{
averageUnionValid : 0.9803530216217
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
Out[25]:
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.9899921298027
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 33 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
Out[25]:
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1022 0 3 0 0 3 0] 99.223% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.99921298027%
+ average rowUcol correct (VOC measure): 98.03530216217%
+ global correct: 99.02%
{
averageUnionValid : 0.9803530216217
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
Out[25]:
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.9899921298027
}
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 34 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1022 0 3 0 0 3 0] 99.223% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.99921298027%
+ average rowUcol correct (VOC measure): 98.03530216217%
+ global correct: 99.02%
{
averageUnionValid : 0.9803530216217
_targ_idx : LongTensor - empty
Out[25]:
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.9899921298027
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 35 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
Out[25]:
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1022 0 3 0 0 3 0] 99.223% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.99921298027%
+ average rowUcol correct (VOC measure): 98.03530216217%
+ global correct: 99.02%
{
averageUnionValid : 0.9803530216217
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.9899921298027
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 36 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
Out[25]:
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1022 0 3 0 0 3 0] 99.223% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.99921298027%
+ average rowUcol correct (VOC measure): 98.03530216217%
+ global correct: 99.02%
{
averageUnionValid : 0.9803530216217
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
Out[25]:
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.9899921298027
}
No Progress Iter: 0
Out[25]:
Backward Progress Iter: 0
on training set:
online epoch # 37 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
Out[25]:
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1022 0 3 0 0 3 0] 99.223% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.99921298027%
+ average rowUcol correct (VOC measure): 98.03530216217%
+ global correct: 99.02%
{
averageUnionValid : 0.9803530216217
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
Out[25]:
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.9899921298027
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 38 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
Out[25]:
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1022 0 3 0 0 3 0] 99.223% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.99921298027%
+ average rowUcol correct (VOC measure): 98.03530216217%
+ global correct: 99.02%
{
averageUnionValid : 0.9803530216217
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.9899921298027
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 39 [batchSize = 100]
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
Out[25]:
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1022 0 3 0 0 3 0] 99.223% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.99921298027%
+ average rowUcol correct (VOC measure): 98.03530216217%
+ global correct: 99.02%
{
averageUnionValid : 0.9803530216217
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.9899921298027
}
Out[25]:
No Progress Iter: 0
Backward Progress Iter: 0
on training set:
online epoch # 40 [batchSize = 100]
Out[25]:
Out[25]:
ConfusionMatrix:
[[ 4929 0 2 0 0 0 0 0 1 0] 99.939% [class: 0]
[ 0 5669 0 0 0 0 1 7 1 0] 99.841% [class: 1]
[ 0 2 4959 0 0 0 2 4 1 0] 99.819% [class: 2]
[ 0 0 1 5092 0 2 0 1 3 2] 99.824% [class: 3]
[ 1 0 0 0 4849 0 3 2 0 4] 99.794% [class: 4]
[ 0 1 1 4 0 4497 3 0 0 0] 99.800% [class: 5]
[ 3 0 1 0 5 1 4939 0 2 0] 99.758% [class: 6]
[ 0 6 7 0 2 0 0 5158 0 2] 99.671% [class: 7]
[ 0 2 2 2 0 4 2 0 4828 2] 99.711% [class: 8]
[ 2 1 0 0 4 2 0 7 2 4970]] 99.639% [class: 9]
+ average row correct: 99.779664874077%
+ average rowUcol correct (VOC measure): 99.561370015144%
+ global correct: 99.78%
{
averageUnionValid : 0.99561370015144
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9978
_prediction : FloatTensor - size: 10
averageValid : 0.99779664874077
}
on testing Set:
Out[25]:
ConfusionMatrix:
[[ 984 0 1 0 0 1 0 1 1 3] 99.294% [class: 0]
[ 0 1060 1 0 0 0 2 1 0 0] 99.624% [class: 1]
[ 1 2 981 2 1 0 1 2 0 0] 99.091% [class: 2]
[ 0 0 2 1022 0 3 0 0 3 0] 99.223% [class: 3]
[ 0 2 0 0 970 0 1 2 1 7] 98.678% [class: 4]
[ 1 0 0 3 2 897 8 0 2 2] 98.033% [class: 5]
[ 0 0 0 0 1 1 964 0 1 0] 99.690% [class: 6]
[ 0 2 0 0 0 0 0 1087 0 1] 99.725% [class: 7]
[ 3 0 1 2 0 1 2 1 995 4] 98.612% [class: 8]
[ 3 0 0 1 7 1 0 4 3 942]] 98.023% [class: 9]
+ average row correct: 98.99921298027%
+ average rowUcol correct (VOC measure): 98.03530216217%
+ global correct: 99.02%
{
averageUnionValid : 0.9803530216217
_targ_idx : LongTensor - empty
valids : FloatTensor - size: 10
classes :
{
1 : 0
2 : 1
3 : 2
4 : 3
5 : 4
6 : 5
7 : 6
8 : 7
9 : 8
10 : 9
}
_target : FloatTensor - empty
mat : FloatTensor - size: 10x10
_pred_idx : LongTensor - size: 1
_max : FloatTensor - size: 1
Out[25]:
unionvalids : FloatTensor - size: 10
nclasses : 10
totalValid : 0.9902
_prediction : FloatTensor - size: 10
averageValid : 0.9899921298027
}
No Progress Iter: 0
Backward Progress Iter: 0
Content source: alexbw/adabayes
Similar notebooks: