Check if tuning hyperparameters match with the model's arguments in the tuning mode

[WenjieDu]

1. Feature description

Check and list mismatched hyperparameters in the error logging file while in the tuning mode.

2. Motivation

While tuning hyperparameters, if there are typos in the searching-space config file, we need to list mismatched arguments, to help users easily identify them.

PyPOTS/pypots/cli/tuning.py

Lines 123 to 147 in 939ad55

	def run(self):
	"""Execute the given command."""
	if os.getenv("enable_tuning", False):
	# fetch the next set of hyperparameters from NNI tuner
	tuner_params = nni.get_next_parameter()
	# get the specified NN class
	model_class = NN_MODELS[self._model]
	# pop out the learning rate
	lr = tuner_params.pop("lr")
	# if tuning a GAN model, we need two optimizers
	if "G_optimizer" in model_class.__init__.__annotations__.keys():
	# optimizer for the generator
	tuner_params["G_optimizer"] = Adam(lr=lr)
	# optimizer for the discriminator
	tuner_params["D_optimizer"] = Adam(lr=lr)
	else:
	tuner_params["optimizer"] = Adam(lr=lr)
	# init an instance with the given hyperparameters for the model class
	model = model_class(**tuner_params)
	# train the model and report to NNI
	model.fit(train_set=self._train_set, val_set=self._val_set)
	else:
	logger.error("Argument `enable_tuning` is not set. Aborting...")

3. Your contribution

Will make a PR to add this feature.