Esempio n. 1
0
 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]
     )
Esempio n. 2
0
 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))