def grid_custom(diz):
import itertools
l = []
lab = [i for i in diz]
for i in diz:
l.append(diz[i])
lines = []
for i in list(itertools.product(*l)):
p = dict(zip(lab, list(i)))
hidden_size = p['hidden_size']
del p['hidden_size']
net = TwoLayerNet(input_size, hidden_size, num_classes)
# Train the network
stats = net.train(X_train, y_train, X_val, y_val, **p)
# Predict on the validation set
val_loss = net.loss(X_val, y_val)[0]
val_acc = (net.predict(X_val) == y_val).mean()
train_loss = net.loss(X_train, y_train)[0]
train_acc = (net.predict(X_train) == y_train).mean()
del p['verbose']
line = []
line.append(hidden_size)
for val in p:
line.append(p[val])
line.append(train_loss)
line.append(val_loss)
line.append(train_acc)
line.append(val_acc)
lines.append(line)
print('params', line)
print('train acc: ', train_acc)
print('test acc: ', val_acc)
pass
df = pd.DataFrame(lines, columns=['hidden_size', 'num_iters', 'batch_size',\
'learning_rate', 'reg', 'train_loss', 'val_loss',\
'train_acc', 'val_acc'])
df.sort_values(by=['val_acc'], ascending=False).to_csv('results_twolayernet.csv', index=False)
return
def init_toy_model():
np.random.seed(0)
return TwoLayerNet(input_size, hidden_size, num_classes, std=1e-1)
plt.imshow(
visualize_grid(X_train[:100, :].reshape(100, 32, 32, 3),
padding=3).astype('uint8'))
plt.gca().axis('off')
plt.show()
# Train a network
# To train our network we will use SGD. In addition, we will
# adjust the learning rate with an exponential learning rate schedule as
# optimization proceeds; after each epoch, we will reduce the learning rate by
# multiplying it by a decay rate.
input_size = 32 * 32 * 3
hidden_size = 50
num_classes = 10
net = TwoLayerNet(input_size, hidden_size, num_classes)
# Train the network
stats = net.train(X_train,
y_train,
X_val,
y_val,
num_iters=1000,
batch_size=200,
learning_rate=1e-4,
learning_rate_decay=0.95,
reg=0.25,
verbose=True)
# Predict on the validation set
val_acc = (net.predict(X_val) == y_val).mean()
print('Validation accuracy: ', val_acc)
# **Approximate results**. You should be aim to achieve a classification
# accuracy of greater than 48% on the validation set. Our best network gets
# over 52% on the validation set.
#
# **Experiment**: You goal in this exercise is to get as good of a result on
# CIFAR-10 as you can (52% could serve as a reference), with a fully-connected
# Neural Network. Feel free implement your own techniques (e.g. PCA to reduce
# dimensionality, or adding dropout, or adding features to the solver, etc.).
# **Explain your hyperparameter tuning process in the report.**
input_size = 32 * 32 * 3
hidden_size = 80
num_classes = 10
# store the best model into this
best_net = TwoLayerNet(input_size, hidden_size, num_classes)
stats = best_net.train(X_train,
y_train,
X_val,
y_val,
num_iters=1000,
batch_size=250,
learning_rate=0.001,
learning_rate_decay=0.99,
reg=0.12,
verbose=True)
# # Predict on the validation set
val_acc = (best_net.predict(X_val) == y_val).mean()
print('Validation accuracy: ', val_acc)
#Plot the loss function and train / validation accuraciess
plt.imshow(
visualize_grid(X_train[:100, :].reshape(100, 32, 32, 3),
padding=3).astype('uint8'))
plt.gca().axis('off')
plt.show()
# Train a network
# To train our network we will use SGD. In addition, we will
# adjust the learning rate with an exponential learning rate schedule as
# optimization proceeds; after each epoch, we will reduce the learning rate by
# multiplying it by a decay rate.
input_size = 32 * 32 * 3
hidden_size = 50
num_classes = 10
net = TwoLayerNet(input_size, hidden_size, num_classes)
# Train the network
stats = net.train(X_train,
y_train,
X_val,
y_val,
num_iters=1000,
batch_size=200,
learning_rate=1e-4,
learning_rate_decay=0.95,
reg=0.25,
verbose=True)
# Predict on the validation set
val_acc = (net.predict(X_val) == y_val).mean()
print('Validation accuracy: ', val_acc)
plt.imshow(
visualize_grid(X_train[:100, :].reshape(100, 32, 32, 3),
padding=3).astype('uint8'))
plt.gca().axis('off')
plt.show()
# Train a network
# To train our network we will use SGD. In addition, we will
# adjust the learning rate with an exponential learning rate schedule as
# optimization proceeds; after each epoch, we will reduce the learning rate by
# multiplying it by a decay rate.
input_size = 32 * 32 * 3
hidden_size = 50
num_classes = 10
net = TwoLayerNet(input_size, hidden_size, num_classes)
# Train the network
stats = net.train(X_train,
y_train,
X_val,
y_val,
num_iters=1000,
batch_size=200,
learning_rate=1e-4,
learning_rate_decay=0.95,
reg=0.25,
verbose=True)
# Predict on the validation set
val_acc = (net.predict(X_val) == y_val).mean()
print('Validation accuracy: ', val_acc)
plt.imshow(
visualize_grid(X_train[:100, :].reshape(100, 32, 32, 3),
padding=3).astype('uint8'))
plt.gca().axis('off')
plt.show()
# Train a network
# To train our network we will use SGD. In addition, we will
# adjust the learning rate with an exponential learning rate schedule as
# optimization proceeds; after each epoch, we will reduce the learning rate by
# multiplying it by a decay rate.
input_size = 32 * 32 * 3
hidden_size = 50
num_classes = 10
net = TwoLayerNet(input_size, hidden_size, num_classes)
# Train the network
stats = net.train(X_train,
y_train,
X_val,
y_val,
num_iters=1000,
batch_size=200,
learning_rate=1e-4,
learning_rate_decay=0.95,
reg=0.25,
verbose=True)
# Predict on the validation set
val_acc = (net.predict(X_val) == y_val).mean()
print('Validation accuracy: ', val_acc)