def create_experiment(experiment_config):
"""Creates a new `Experiment` instance.
Args:
experiment_config: the config to use for creating the experiment.
"""
# Creates training input function
train_input_data_config = experiment_config.train_input_data_config
train_input_fn = create_input_data_fn(
pipeline_config=train_input_data_config.pipeline_config,
mode=ModeKeys.TRAIN,
scope='train_input_fn',
input_type=train_input_data_config.input_type,
x=train_input_data_config.x,
y=train_input_data_config.y)
# Creates eval_input_fn input function
eval_input_data_config = experiment_config.eval_input_data_config
eval_input_fn = create_input_data_fn(
pipeline_config=eval_input_data_config.pipeline_config,
mode=ModeKeys.EVAL,
scope='eval_input_fn',
input_type=eval_input_data_config.input_type,
x=eval_input_data_config.x,
y=eval_input_data_config.y)
estimator = getters.get_estimator(experiment_config.estimator_config,
experiment_config.model_config,
experiment_config.run_config)
train_hooks = getters.get_hooks(experiment_config.train_hooks_config)
eval_hooks = getters.get_hooks(experiment_config.eval_hooks_config)
experiment = Experiment(
estimator=estimator,
train_input_fn=train_input_fn,
eval_input_fn=eval_input_fn,
train_steps=experiment_config.train_steps,
eval_steps=experiment_config.eval_steps,
train_hooks=train_hooks,
eval_hooks=eval_hooks,
eval_delay_secs=experiment_config.eval_delay_secs,
continuous_eval_throttle_secs=experiment_config.
continuous_eval_throttle_secs,
eval_every_n_steps=experiment_config.eval_every_n_steps,
delay_workers_by_global_step=experiment_config.
delay_workers_by_global_step,
export_strategies=experiment_config.export_strategies,
train_steps_per_iteration=experiment_config.train_steps_per_iteration)
return experiment
def _get_train(config):
assert isinstance(
config, TrainConfig), '`config` must be a an instance of `TrainConfig`'
input_fn = create_input_data_fn(mode=Modes.TRAIN,
pipeline_config=config.data_pipeline)
return input_fn, config.steps, config.hooks
def _get_eval(config):
assert isinstance(
config, EvalConfig), '`config` must be a an instance of `EvalConfig`'
input_fn = create_input_data_fn(mode=Modes.EVAL,
pipeline_config=config.data_pipeline)
return (input_fn, config.steps, config.hooks, config.delay_secs,
config.continuous_eval_throttle_secs)
def create_experiment(experiment_config):
"""Creates a new `Experiment` instance.
Args:
experiment_config: the config to use for creating the experiment.
"""
# Creates training input function
train_input_data_config = experiment_config.train_input_data_config
train_input_fn = create_input_data_fn(
pipeline_config=train_input_data_config.pipeline_config,
mode=Modes.TRAIN, scope='train_input_fn',
input_type=train_input_data_config.input_type,
x=train_input_data_config.x, y=train_input_data_config.y)
# Creates eval_input_fn input function
eval_input_data_config = experiment_config.eval_input_data_config
eval_input_fn = create_input_data_fn(
pipeline_config=eval_input_data_config.pipeline_config,
mode=Modes.EVAL, scope='eval_input_fn',
input_type=eval_input_data_config.input_type,
x=eval_input_data_config.x, y=eval_input_data_config.y)
estimator = getters.get_estimator(experiment_config.estimator_config,
experiment_config.model_config,
experiment_config.run_config)
train_hooks = getters.get_hooks(experiment_config.train_hooks_config)
eval_hooks = getters.get_hooks(experiment_config.eval_hooks_config)
experiment = Experiment(
estimator=estimator,
train_input_fn=train_input_fn,
eval_input_fn=eval_input_fn,
train_steps=experiment_config.train_steps,
eval_steps=experiment_config.eval_steps,
train_hooks=train_hooks,
eval_hooks=eval_hooks,
eval_delay_secs=experiment_config.eval_delay_secs,
continuous_eval_throttle_secs=experiment_config.continuous_eval_throttle_secs,
eval_every_n_steps=experiment_config.eval_every_n_steps,
delay_workers_by_global_step=experiment_config.delay_workers_by_global_step,
export_strategies=experiment_config.export_strategies,
train_steps_per_iteration=experiment_config.train_steps_per_iteration)
return experiment