def train_model(dataset, h0_dim, h1_dim, out_dim):
X_train, y_train, X_test, y_test = dataset
batch_size = 128
nb_epoch = 100
model = Sequential()
model.add(
RNN(h0_dim,
input_shape=(None, X_train.shape[-1]),
return_sequences=True))
model.add(TimeDistributedDense(out_dim))
model.add(Activation("linear"))
model.compile(loss="mse", optimizer="rmsprop")
#model.get_config(verbose=1)
#yaml_string = model.to_yaml()
#with open('ifshort_mlp.yaml', 'w') as f:
# f.write(yaml_string)
early_stopping = EarlyStopping(monitor='val_loss', patience=10)
checkpointer = ModelCheckpoint(filepath="/tmp/ifshort_rnn_weights.hdf5",
verbose=1,
save_best_only=True)
model.fit(X_train,
y_train,
batch_size=batch_size,
nb_epoch=nb_epoch,
show_accuracy=False,
verbose=2,
validation_data=(X_test, y_test),
callbacks=[early_stopping, checkpointer])
def train_model(dataset, h0_dim, h1_dim, y_dim):
X_train, y_train, X_test, y_test = dataset
batch_size = 512
nb_epoch = 100
model = Sequential()
model.add(Dense(h0_dim, input_shape=(X_train.shape[1],),
init='uniform',
W_regularizer=l2(0.0005),
activation='relu'))
model.add(Dense(h1_dim,
init='uniform',
W_regularizer=l2(0.0005),
activation='relu'))
model.add(Dense(y_dim,
init='uniform',
W_regularizer=l2(0.0005)))
rms = RMSprop()
sgd = SGD(lr=0.01, decay=1e-4, momentum=0.6, nesterov=False)
model.compile(loss='mse', optimizer=sgd)
#model.get_config(verbose=1)
#yaml_string = model.to_yaml()
#with open('ifshort_mlp.yaml', 'w') as f:
# f.write(yaml_string)
early_stopping = EarlyStopping(monitor='val_loss', patience=10)
checkpointer = ModelCheckpoint(filepath="/tmp/ifshort_mlp_weights.hdf5", verbose=1, save_best_only=True)
model.fit(X_train, y_train,
batch_size=batch_size,
nb_epoch=nb_epoch,
show_accuracy=False,
verbose=2,
validation_data=(X_test, y_test),
callbacks=[early_stopping, checkpointer])
def train_model(dataset, h0_dim, h1_dim, out_dim):
X_train, y_train, X_test, y_test = dataset
batch_size = 128
nb_epoch = 100
model = Sequential()
model.add(RNN(h0_dim, input_shape=(None, X_train.shape[-1]), return_sequences=True))
model.add(TimeDistributedDense(out_dim))
model.add(Activation("linear"))
model.compile(loss="mse", optimizer="rmsprop")
#model.get_config(verbose=1)
#yaml_string = model.to_yaml()
#with open('ifshort_mlp.yaml', 'w') as f:
# f.write(yaml_string)
early_stopping = EarlyStopping(monitor='val_loss', patience=10)
checkpointer = ModelCheckpoint(filepath="/tmp/ifshort_rnn_weights.hdf5", verbose=1, save_best_only=True)
model.fit(X_train, y_train,
batch_size=batch_size,
nb_epoch=nb_epoch,
show_accuracy=False,
verbose=2,
validation_data=(X_test, y_test),
callbacks=[early_stopping, checkpointer])
activation='relu'))
#model.add(Dropout(0.2))
model.add(Dense(40,
init='uniform',
# init=lambda shape: uniform(shape, scale=0.05),
W_regularizer=l2(0.0005),
activation='relu'))
#model.add(Dropout(0.2))
model.add(Dense(1,
init='uniform',
# init=lambda shape: uniform(shape, scale=0.005),
W_regularizer=l2(0.0005)))
rms = RMSprop()
sgd = SGD(lr=0.01, decay=1e-4, momentum=0.6, nesterov=False)
model.compile(loss='mse', optimizer=sgd)
#model.get_config(verbose=1)
yaml_string = model.to_yaml()
with open('ifshort_mlp.yaml', 'w') as f:
f.write(yaml_string)
early_stopping = EarlyStopping(monitor='val_loss', patience=10)
checkpointer = ModelCheckpoint(filepath="/tmp/ifshort_mlp_weights.hdf5", verbose=1, save_best_only=True)
model.fit(X_train, y_train,
batch_size=batch_size,
nb_epoch=nb_epoch,
show_accuracy=False,
verbose=2,
validation_data=(X_test, y_test),
callbacks=[early_stopping, checkpointer])
W_regularizer=l2(0.0005),
activation='relu'))
#model.add(Dropout(0.2))
model.add(
Dense(
1,
init='uniform',
# init=lambda shape: uniform(shape, scale=0.005),
W_regularizer=l2(0.0005)))
rms = RMSprop()
sgd = SGD(lr=0.01, decay=1e-4, momentum=0.6, nesterov=False)
model.compile(loss='mse', optimizer=sgd)
#model.get_config(verbose=1)
yaml_string = model.to_yaml()
with open('ifshort_mlp.yaml', 'w') as f:
f.write(yaml_string)
early_stopping = EarlyStopping(monitor='val_loss', patience=10)
checkpointer = ModelCheckpoint(filepath="/tmp/ifshort_mlp_weights.hdf5",
verbose=1,
save_best_only=True)
model.fit(X_train,
y_train,
batch_size=batch_size,
nb_epoch=nb_epoch,
show_accuracy=False,
verbose=2,
validation_data=(X_test, y_test),
callbacks=[early_stopping, checkpointer])
