def _prepare_for_class(self, inputs_dict, model_class, return_labels=False) -> dict:
inputs_dict = copy.deepcopy(inputs_dict)
if model_class in TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING.values():
inputs_dict = {
k: tf.tile(tf.expand_dims(v, 1), (1, self.model_tester.num_choices) + (1,) * (v.ndim - 1))
if isinstance(v, tf.Tensor) and v.ndim > 0
else v
for k, v in inputs_dict.items()
}
if return_labels:
if model_class in TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING.values():
inputs_dict["labels"] = tf.ones(self.model_tester.batch_size, dtype=tf.int32)
elif model_class in TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING.values():
inputs_dict["start_positions"] = tf.zeros(self.model_tester.batch_size, dtype=tf.int32)
inputs_dict["end_positions"] = tf.zeros(self.model_tester.batch_size, dtype=tf.int32)
elif model_class in TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING.values():
inputs_dict["labels"] = tf.zeros(self.model_tester.batch_size, dtype=tf.int32)
elif model_class in TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING.values():
inputs_dict["next_sentence_label"] = tf.zeros(self.model_tester.batch_size, dtype=tf.int32)
elif model_class in [
*TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING.values(),
*TF_MODEL_FOR_CAUSAL_LM_MAPPING.values(),
*TF_MODEL_FOR_MASKED_LM_MAPPING.values(),
*TF_MODEL_FOR_PRETRAINING_MAPPING.values(),
*TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING.values(),
]:
inputs_dict["labels"] = tf.zeros(
(self.model_tester.batch_size, self.model_tester.seq_length), dtype=tf.int32
)
return inputs_dict
def _prepare_for_class(self, inputs_dict, model_class, return_labels=False):
inputs_dict = super()._prepare_for_class(inputs_dict, model_class, return_labels=return_labels)
if return_labels:
if model_class in TF_MODEL_FOR_PRETRAINING_MAPPING.values():
inputs_dict["next_sentence_label"] = tf.zeros(self.model_tester.batch_size, dtype=tf.int32)
return inputs_dict