def create_model():
"""Factory method that creates model to be used by generic task.py."""
parser = argparse.ArgumentParser()
# Label count needs to correspond to nubmer of labels in dictionary used
# during preprocessing.
parser.add_argument('--label_count', type=int, default=LABEL_COUNT)
parser.add_argument('--dropout', type=float, default=0.5)
parser.add_argument(
'--inception_checkpoint_file',
type=str,
default=DEFAULT_INCEPTION_CHECKPOINT)
args, task_args = parser.parse_known_args()
# Adding the following 'hack' to make the example easier to run: use the
# classifier_label count if it was specified as a task.py arg, to set the
# model label_count.
if '--classifier_label_count' in task_args:
clabel_count = task_args[task_args.index('--classifier_label_count') + 1]
logging.info("Found classifier_label_count task arg.")
try:
clabel_count = int(clabel_count)
args.label_count = clabel_count
except:
logging.info("classifier label count was set, but is not an int.")
logging.info("using label count: %s", args.label_count)
override_if_not_in_args('--max_steps', '1000', task_args)
override_if_not_in_args('--batch_size', '100', task_args)
override_if_not_in_args('--eval_set_size', '370', task_args)
override_if_not_in_args('--eval_interval_secs', '2', task_args)
override_if_not_in_args('--log_interval_secs', '2', task_args)
override_if_not_in_args('--min_train_eval_rate', '2', task_args)
return Model(args.label_count, args.dropout,
args.inception_checkpoint_file), task_args
def create_model():
"""Factory method that creates model to be used by generic task.py."""
parser = argparse.ArgumentParser()
parser.add_argument('--learning_rate', type=float, default=0.01)
args, task_args = parser.parse_known_args()
override_if_not_in_args('--max_steps', '5000', task_args)
override_if_not_in_args('--batch_size', '100', task_args)
override_if_not_in_args('--eval_set_size', '10000', task_args)
override_if_not_in_args('--eval_interval_secs', '1', task_args)
override_if_not_in_args('--log_interval_secs', '1', task_args)
override_if_not_in_args('--min_train_eval_rate', '1', task_args)
return Model(args.learning_rate, HIDDEN1, HIDDEN2), task_args
def create_model():
"""Factory method that creates model to be used by generic task.py."""
parser = argparse.ArgumentParser()
parser.add_argument('--learning_rate', type=float, default=0.002)
args, task_args = parser.parse_known_args()
override_if_not_in_args('--max_steps', '5000', task_args)
override_if_not_in_args('--batch_size', '256', task_args)
override_if_not_in_args('--eval_set_size', '10000', task_args)
override_if_not_in_args('--eval_interval_secs', '1', task_args)
override_if_not_in_args('--log_interval_secs', '1', task_args)
override_if_not_in_args('--min_train_eval_rate', '1', task_args)
return Model(args.learning_rate, HIDDEN1, HIDDEN2), task_args
def create_model():
"""Factory method that creates model to be used by generic task.py."""
# HYPERPARAMETER TUNING: These flags are set by Cloud ML from the
# hyperparameters defined in the API call. They will be passed in as normal
# command line flags.
parser = argparse.ArgumentParser()
parser.add_argument('--learning_rate', type=float, default=0.01)
parser.add_argument('--hidden1', type=int, default=128)
parser.add_argument('--hidden2', type=int, default=32)
args, task_args = parser.parse_known_args()
override_if_not_in_args('--max_steps', '5000', task_args)
override_if_not_in_args('--batch_size', '100', task_args)
override_if_not_in_args('--eval_set_size', '10000', task_args)
# HYPERPARAMETER TUNING: Do not write the objective value too frequently.
override_if_not_in_args('--eval_interval_secs', '10', task_args)
override_if_not_in_args('--log_interval_secs', '10', task_args)
override_if_not_in_args('--min_train_eval_rate', '5', task_args)
return Model(args.learning_rate, args.hidden1, args.hidden2), task_args
def create_model():
"""Factory method that creates model to be used by generic task.py."""
parser = argparse.ArgumentParser()
# Label count needs to correspond to nubmer of labels in dictionary used
# during preprocessing.
parser.add_argument('--label_count', type=int, default=5)
parser.add_argument('--dropout', type=float, default=0.5)
parser.add_argument(
'--inception_checkpoint_file',
type=str,
default=DEFAULT_INCEPTION_CHECKPOINT)
args, task_args = parser.parse_known_args()
override_if_not_in_args('--max_steps', '1000', task_args)
override_if_not_in_args('--batch_size', '100', task_args)
override_if_not_in_args('--eval_set_size', '370', task_args)
override_if_not_in_args('--eval_interval_secs', '2', task_args)
override_if_not_in_args('--log_interval_secs', '2', task_args)
override_if_not_in_args('--min_train_eval_rate', '2', task_args)
return Model(args.label_count, args.dropout,
args.inception_checkpoint_file), task_args
