def create_and_check_roberta_for_token_classification(
self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
):
config.num_labels = self.num_labels
model = TFRobertaForTokenClassification(config=config)
inputs = {"input_ids": input_ids, "attention_mask": input_mask, "token_type_ids": token_type_ids}
(logits,) = model(inputs)
result = {
"logits": logits.numpy(),
}
self.parent.assertListEqual(
list(result["logits"].shape), [self.batch_size, self.seq_length, self.num_labels]
)
def create_and_check_roberta_for_token_classification(
self, config, input_ids, token_type_ids, input_mask,
sequence_labels, token_labels, choice_labels):
config.num_labels = self.num_labels
model = TFRobertaForTokenClassification(config=config)
inputs = {
"input_ids": input_ids,
"attention_mask": input_mask,
"token_type_ids": token_type_ids
}
result = model(inputs)
self.parent.assertEqual(
result.logits.shape,
(self.batch_size, self.seq_length, self.num_labels))