overfit_weight_decay.py

from dataset import load_mnist
from common.multi_layer_net import MultiLayerNet
from common.optimizer import SGD
import matplotlib.pyplot as plt
import numpy as np
import tqdm
(x_train, t_train), (x_test, t_test) = load_mnist(normolize=True,
                                                  one_hot_lable=True)
x_train = x_train[:300]
t_train = t_train[:300]

network = MultiLayerNet(input_size=784,
                        hidden_size_list=[100, 100, 100, 100, 100, 100],
                        output_size=10)

opt = SGD(lr=0.01)

max_epochs = 201
train_size = x_train.shape[0]
batch_size = 100

trian_loss_list = []
train_acc_list = []
test_acc_list = []

iter_per_epoch = max(train_size / batch_size, 1)

epoch_cnt = 0

for i in tqdm.tqdm(range(10000)):
    batch_mask = np.random.choice(train_size, batch_size)

    x_batch = x_train[batch_mask]
    t_batch = t_train[batch_mask]

    grads = network.gradient(x_batch, t_batch)
    opt.update(network.params, grads)

    if i % iter_per_epoch == 0:
        train_acc = network.accuracy(x_train, t_train)
        test_acc = network.accuracy(x_test, t_test)
        train_acc_list.append(train_acc)
        test_acc_list.append(test_acc)

        epoch_cnt += 1

        if epoch_cnt >= max_epochs:
            break
x = np.arange(max_epochs)
y1 = np.array(train_acc_list)
y2 = np.array(test_acc_list)

plt.figure()
plt.plot(x, y1, label="train_acc")
plt.plot(x, y2, label="test_acc")
plt.legend()
plt.show()