GitHub - hillbig/binary_net: experimental binary net implementation in chainer

load MNIST dataset epoch 1 graph generated train mean loss=0.573861178756, accuracy=0.92756666926 test mean loss=0.473955234885, accuracy=0.957400003672 epoch 2 train mean loss=0.456328810602, accuracy=0.963833337426 test mean loss=0.436628208458, accuracy=0.966100006104 epoch 3 train mean loss=0.431186137001, accuracy=0.970866675178 test mean loss=0.425710965991, accuracy=0.968000004292 epoch 4 t

[kitanokumo]

binarynetのmnist 精度はオリジナルmnistサンプルより若干劣る 実行速度も