def main(argv):
batch_size = 200
epochs = 50
# Load MNIST dataset
(x_train, y_train), (x_test, y_test) = mnist.load_data()
# Convert images from [0,255] to [0,1]
x_train = x_train.astype('float32') / 255.0
x_test = x_test.astype('float32') / 255.0
# Extract input dimension and distinct output classes from dataset
input_shape = x_train.shape[1:]
num_classes = len(np.unique(y_train))
num_train_samples = x_train.shape[0]
# Convert y to categorical one-hot vectors
y_train = to_categorical(y_train, num_classes=num_classes)
y_test = to_categorical(y_test, num_classes=num_classes)
# Create and compile model
model = mnist_dense_model(input_shape=input_shape,
n_classes=num_classes,
dropout=0.5,
model_name=get_project_name(argv[0]))
model.compile(loss=categorical_crossentropy,
optimizer=Adam(),
metrics=['accuracy'])
# Print summary and save model as plot and node-link-graph
project_paths = get_project_paths(argv[0], to_tmp=False)
save_graph_plot(model, project_paths["plots"] + "/model.ps")
save_graph_json(model, project_paths["graphs"] + "/model.json")
import pdb
pdb.set_trace()
# Train model while saving weights as checkpoints after each epoch
model.fit(
x_train,
y_train,
steps_per_epoch=num_train_samples / batch_size /
5, # 5 epochs per full dataset rotation
batch_size=batch_size,
epochs=epochs,
verbose=1,
callbacks=[
ModelCheckpoint(project_paths["checkpoints"] +
"/weights_epoch-{epoch:02d}.hdf5",
save_weights_only=True,
save_freq='epoch')
],
validation_data=(x_test, y_test))
keras.Input(shape=input_shape),
layers.Conv2D(2, kernel_size=(3, 3), activation="relu"),
layers.MaxPooling2D(pool_size=(2, 2)),
layers.Conv2D(4, kernel_size=(3, 3), activation="relu"),
layers.MaxPooling2D(pool_size=(2, 2)),
layers.Flatten(),
layers.Dropout(0.5),
layers.Dense(num_classes, activation="softmax"),
])
return model
model1 = create_model()
model1.summary()
save_graph_plot(model1, project_paths["plots"] + "/reg_model.ps")
save_graph_json(model1, project_paths["weights"] + "/reg_model.json")
date_time = datetime.datetime.now().strftime("%Y%m%d-%H%M%S")
log_dir = project_paths["tb"] + "/reg/" + date_time
tensorboard_callback1 = tf.keras.callbacks.TensorBoard(log_dir=log_dir,
histogram_freq=1)
checkpoint_path = project_paths["checkpoints"] + "/weights_epoch-{epoch}.ckpt"
checkpoint_dir = os.path.dirname(checkpoint_path)
"""
## Train the model
"""
batch_size = 128
model1.compile(loss="categorical_crossentropy",
optimizer="adam",
metrics=["accuracy"])
def create_model():
model = models.Sequential()
model.add(layers.Flatten(input_shape=(28, 28)))
model.add(layers.Dense(2, activation='relu'))
model.add(layers.Dense(10))
model.compile(
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'],
optimizer='adam')
return model
reg_model = create_model()
save_graph_plot(reg_model, project_paths["plots"] + "/reg_model.ps")
save_graph_json(reg_model, project_paths["plots"] + "/reg_model.json")
reg_model_hist = reg_model.fit(
train_images,
train_labels,
epochs=epochs,
validation_data=(test_images, test_labels),
callbacks=[csv_logger, callback_save_model_reg, callback_weights_reg])
model_list.append(reg_model_hist)
model_name_list.append("Regular model ")
# Pred Model
# pred_model = create_model()
pred_model = tf.keras.models.load_model(restore_path)
model.add(tf.keras.layers.MaxPooling2D((2, 2)))
model.add(tf.keras.layers.Conv2D(128, (3, 3), activation='relu'))
model.add(tf.keras.layers.MaxPooling2D(pool_size=(2, 2)))
model.add(layers.Flatten())
model.add(layers.Dense(256, activation='relu'))
model.add(layers.Dense(128, activation='relu'))
model.add(layers.Dense(num_classes))
model.summary()
model.compile(
optimizer='adam',
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])
project_paths = get_project_paths(sys.argv[0], to_tmp=False)
save_graph_plot(model, project_paths["plots"] + "/model.ps")
save_graph_json(model, project_paths["weights"] + "/model.json")
log_dir = "logs/fit/" + datetime.datetime.now().strftime("%Y%m%d-%H%M%S")
tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir=log_dir,
histogram_freq=1)
history = model.fit(
train_images,
train_labels,
steps_per_epoch=num_train_samples /
batch_size, # 5 epochs per full dataset rotation
epochs=epochs,
batch_size=batch_size,
#validation_split=validation_split,
validation_data=(test_images, test_labels),
callbacks=[
