test_acc_list = []
model = NeuralNet()
model.add_affine(784, 50)
model.add_active("relu")
model.add_affine(50, 10)
for i in range(iters_num):
batch_mask = np.random.choice(train_size, batch_size)
x_batch = x_train[batch_mask]
t_batch = t_train[batch_mask]
model.fit(x_batch, t_batch, learning_rate)
loss = model.loss(x_batch, t_batch)
train_loss_list.append(loss)
if i % iter_per_epoch == 0:
train_acc = model.accuracy(x_train, t_train)
test_acc = model.accuracy(x_test, t_test)
train_acc_list.append(train_acc)
test_acc_list.append(test_acc)
print(str(i) + "回目")
print(train_acc, test_acc)
x = np.arange(len(train_loss_list))
plt.plot(x, train_loss_list)
plt.show()
