新マシン GeForce GTX 1080 Ti その２

Chainer 3.2.0 で MNIST の実行を行ってみました。

time python ./train_mnist.py -g 1
GPU: 1
# unit: 1000
# Minibatch-size: 100
# epoch: 20

epoch main/loss validation/main/loss main/accuracy validation/main/accuracy elapsed_time
1 0.19217 0.0908331 0.941234 0.9706 4.18323
2 0.0728654 0.0949544 0.976833 0.9705 7.54424
3 0.0468418 0.0857347 0.984999 0.9733 10.9104
4 0.0367249 0.0692376 0.988115 0.9795 14.2286
5 0.029575 0.0858509 0.990298 0.9769 17.6992
6 0.0225167 0.0738301 0.992498 0.9801 21.2803
7 0.022077 0.0807188 0.992532 0.9801 25.1095
8 0.0194491 0.0746321 0.993948 0.9812 28.5453
9 0.0125162 0.083389 0.995782 0.9833 31.859
10 0.0157655 0.0925374 0.995082 0.9782 35.1737
11 0.0156083 0.0888666 0.995299 0.981 38.469
12 0.0131244 0.10147 0.995799 0.9801 41.7732
13 0.0105811 0.0843209 0.996365 0.983 45.0925
14 0.0128965 0.0911568 0.996099 0.9806 48.4146
15 0.00683104 0.105298 0.997882 0.9802 51.7209
16 0.010362 0.102377 0.996932 0.9834 55.0434
17 0.0131538 0.0922663 0.996032 0.9823 58.8243
18 0.00833331 0.107601 0.997382 0.9815 62.45
19 0.0120943 0.110463 0.996666 0.9822 66.1032
20 0.00927191 0.126229 0.997482 0.9805 69.7293

real 1m16.226s
user 1m13.981s
sys 0m18.811s