def run(self):
"""
start inferring
:return:
"""
# prepare data
self.eval_data = self.prepare_data()
# split data
x_eval, y_eval = self.eval_data
# init configs from checkpoints json file and flags
config = load_config(self.flags)
# init model class
model_class_name, model_file = config['model_class'], config[
'model_file']
self.model_class = find_model(model_class_name, model_file)
# init model
model = self.model_class(config)
model.init()
# init variables
model.set_inputs(x_eval)
# restore model
load_model(model, self.flags.checkpoint_dir,
self.flags.checkpoint_name)
# evaluate
return model.evaluate(x_eval, y_eval)
def run(self, **kwargs):
"""
start inferring
:return:
"""
# prepare data
self.eval_data = self.data()
# split data
x_eval, y_eval = self.eval_data
# init configs from checkpoints json file and flags
config = load_config(self.config)
# init model class
model_class_name, model_file_name = config.get(
'model_class_name'), config.get('model_file_name')
self.model_class = find_model_class(model_class_name, model_file_name)
# init model
model = self.model_class(config=config)
model.logger = self.logger
self.logger.info(f'initialize model logger {model.logger} of {model}')
# restore model
load_model(model, self.config.get('checkpoint_dir'),
self.config.get('checkpoint_name'))
# evaluate
return model.evaluate(x_eval, y_eval, **kwargs)
def run(self):
"""
start inferring
:return:
"""
# get test_data
self.test_data = self.prepare_data()
# init model
model = self.model_class(load_config(self.flags))
model.init()
# init variables
model.call(tf.keras.Input(shape=(get_shape(self.test_data))),
training=False)
# restore model if exists
load_model(model, self.flags.checkpoint_dir,
self.flags.checkpoint_name)
# infer
return model.infer(self.test_data)
def run(self):
"""
start inferring
:return:
"""
# prepare data
self.eval_data = self.prepare_data()
# split data
x_eval, y_eval = self.eval_data
# init model
model = self.model_class(load_config(self.flags))
model.init()
# init variables
model.call(tf.keras.Input(shape=(get_shape(x_eval))), training=False)
# restore model
load_model(model, self.flags.checkpoint_dir,
self.flags.checkpoint_name)
# evaluate
return model.evaluate(x_eval, y_eval)
def run(self, **kwargs):
"""
start inferring
:return:
"""
# get test_data
self.test_data = self.data()
# init configs from checkpoints json file and flags
config = load_config(self.config)
# init model class
model_class_name, model_file_name = config.get(
'model_class_name'), config.get('model_file_name')
self.model_class = find_model_class(model_class_name, model_file_name)
# init model
model = self.model_class(config=config)
# restore model if exists
load_model(model, self.config.get('checkpoint_dir'),
self.config.get('checkpoint_name'))
# infer
return model.infer(self.test_data, **kwargs)
def run(self):
"""
start inferring
:return:
"""
# get test_data
self.test_data = self.prepare_data()
# init configs from checkpoints json file and flags
config = load_config(self.flags)
# init model class
model_class_name, model_file = config['model_class'], config[
'model_file']
self.model_class = find_model(model_class_name, model_file)
# init model
model = self.model_class(config)
model.init()
# init variables
model.set_inputs(self.test_data)
# restore model if exists
load_model(model, self.flags.checkpoint_dir,
self.flags.checkpoint_name)
# infer
return model.infer(self.test_data)