lr_rate = 0.0001
batch_size = 16
# Step 1: Create Model
model = conv_ae.Conv_AE((None, height, width, channel), latent=200, units=16)
if sys.argv[1] == "train":
print(model.summary())
# sys.exit()
# Load weights:
# model.load_weights(model_name)
# Step 3: Load data
X_train, Y_train, X_valid, Y_valid = loader.load_light(
data_path, width, height, True, 0.8, False)
# Define The Optimizer
optimizer = tf.keras.optimizers.Adam(learning_rate=lr_rate)
# Define The Loss
#---------------------
@tf.function
def my_loss(y_true, y_pred):
return tf.keras.losses.MSE(y_true=y_true, y_pred=y_pred)
# Define The Metrics
tr_loss = tf.keras.metrics.MeanSquaredError(name='tr_loss')
va_loss = tf.keras.metrics.MeanSquaredError(name='va_loss')
#---------------------
@tf.function
def train_step(X, Y_true):
lr_rate = 1e-4
batch_size = 4
# Step 1: Create Model
model = conv_vae.CONV_VAE(image_size=image_size,
latent_dim=latent_dim,
filters=6)
model.build((None, image_size, image_size, 3))
# Step 2: Define Metrics
# print(model.summary())
# sys.exit()
if sys.argv[1] == "train":
# Step 3: Load data
X_train, Y_train, X_valid, Y_valid = loader.load_light(
data_path, image_size, image_size, True, 0.8, True)
# Step 4: Training
# model.load_weights(model_name)
# Define The Optimizer
optimizer = tf.keras.optimizers.Adam(
learning_rate=lr_rate) #, beta_1 = 0.5)
@tf.function
def ae_loss(y_true, y_pred):
# de_loss = tf.reduce_mean(tf.pow(y_true - y_pred, 2))
return tf.keras.losses.MSE(y_true=y_true, y_pred=y_pred)
# Define The Metrics
tr_loss = tf.keras.metrics.MeanSquaredError(name='tr_loss')