未归一化的比较： training on cpu epoch 1, loss 0.0091, train acc 0.104, test acc 0.100, time 61.8 sec epoch 2, loss 0.0053, train acc 0.464, test acc 0.611, time 59.1 sec epoch 3, loss 0.0034, train acc 0.652, test acc 0.557, time 61.1 sec epoch 4, loss 0.0029, train acc 0.710, test acc 0.645, time 58.6 sec epoch 5, loss 0.0025, train acc 0.750, test acc 0.732, time 58.9 sec epoch 6, loss 0.0023, train acc 0.776, test acc 0.778, time 58.1 sec epoch 7, loss 0.0021, train acc 0.794, test acc 0.801, time 58.4 sec epoch 8, loss 0.0019, train acc 0.812, test acc 0.778, time 58.1 sec epoch 9, loss 0.0018, train acc 0.823, test acc 0.815, time 61.2 sec epoch 10, loss 0.0018, train acc 0.832, test acc 0.823, time 68.8 sec
批量归一化：
training on cpu epoch 1, loss 0.0026, train acc 0.774, test acc 0.649, time 161.5 sec epoch 2, loss 0.0015, train acc 0.865, test acc 0.789, time 157.5 sec epoch 3, loss 0.0013, train acc 0.882, test acc 0.819, time 156.7 sec epoch 4, loss 0.0012, train acc 0.891, test acc 0.644, time 161.3 sec epoch 5, loss 0.0011, train acc 0.898, test acc 0.844, time 168.6 sec epoch 6, loss 0.0010, train acc 0.903, test acc 0.819, time 174.6 sec epoch 7, loss 0.0010, train acc 0.907, test acc 0.792, time 163.0 sec epoch 8, loss 0.0010, train acc 0.912, test acc 0.758, time 178.3 sec epoch 9, loss 0.0009, train acc 0.914, test acc 0.869, time 174.0 sec epoch 10, loss 0.0009, train acc 0.917, test acc 0.840, time 177.8 sec