def test_tf_predict(self):
Mnist(self.mnist_dir, mode='test').b(DATA_BRANCH).run()
TFPredict(cluster_size=3,
num_ps=1,
steps=10,
params='{"keep_prob": 1}',
model_dir=self.model_dir).run()
def test_evaluate_model(self):
Mnist(self.mnist_dir, mode='test').b(DATA_BRANCH).run()
EvaluateModel(input_prev_layers=MODEL_BRANCH,
input_rdd_name=DATA_BRANCH,
cluster_size=3,
num_ps=1,
steps=0,
model_dir=self.model_dir).run()
def test_predict_model(self):
Mnist(self.mnist_dir, mode='test').b(DATA_BRANCH).run()
PredictModel(input_prev_layers=MODEL_BRANCH,
input_rdd_name=DATA_BRANCH,
cluster_size=3,
num_ps=1,
steps=10,
model_dir=self.model_dir,
output_prob='true').run()
def test_current_predict_model(self):
Mnist(self.mnist_dir, mode='test').b(DATA_BRANCH).run()
RecurrentPredictModel(input_prev_layers=MODEL_BRANCH,
input_rdd_name=DATA_BRANCH,
cluster_size=3,
num_ps=1,
units=784,
steps=10,
feature_type='one_hot',
model_dir=self.model_dir).run()
def test_lenet_evaluate(self):
# load train data
Mnist(self.mnist_dir, mode='test', is_conv='true').b(DATA_BRANCH).run()
# model train
EvaluateModel(input_prev_layers=MODEL_BRANCH,
input_rdd_name=DATA_BRANCH,
cluster_size=2,
num_ps=1,
steps=0,
model_dir=self.model_dir).run()
def test_train_model(self):
Mnist(self.mnist_dir, mode='train').b(DATA_BRANCH).run()
self.build_model()
TrainModel(input_prev_layers=MODEL_BRANCH,
input_rdd_name=DATA_BRANCH,
cluster_size=3,
num_ps=1,
batch_size=32,
epochs=2,
model_dir=self.model_dir,
go_on='false').run()
def test_lenet_predict(self):
# load train data
Mnist(self.mnist_dir, mode='test', is_conv='true').b(DATA_BRANCH).run()
# model predict
PredictModel(input_prev_layers=MODEL_BRANCH,
input_rdd_name=DATA_BRANCH,
cluster_size=2,
num_ps=1,
steps=10,
model_dir=self.model_dir,
output_prob='true').run()
def test_tf_train(self):
Mnist(self.mnist_dir, mode='train').b(DATA_BRANCH).run()
MLPModel().b(MODEL_BRANCH).run()
MLPCompile().run()
TFTrain(input_prev_layers=MODEL_BRANCH,
input_rdd_name=DATA_BRANCH,
cluster_size=3,
num_ps=1,
batch_size=32,
epochs=2,
model_dir=self.model_dir,
go_on='false').run()
def test_alexnet_train(self):
# load train data
Mnist(self.mnist_dir, mode='train', is_conv='true').b(DATA_BRANCH).run()
self.build_model()
# model train
TrainModel(input_prev_layers=MODEL_BRANCH,
input_rdd_name=DATA_BRANCH,
cluster_size=3,
num_ps=1,
batch_size=32,
epochs=2,
model_dir=self.model_dir,
go_on='false').run()