def create_and_check_albert_for_pretraining( self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels ): config.num_labels = self.num_labels model = TFAlbertForPreTraining(config=config) inputs = {"input_ids": input_ids, "attention_mask": input_mask, "token_type_ids": token_type_ids} result = model(inputs) self.parent.assertEqual(result.prediction_logits.shape, (self.batch_size, self.seq_length, self.vocab_size)) self.parent.assertEqual(result.sop_logits.shape, (self.batch_size, self.num_labels))
def test_inference_masked_lm(self): model = TFAlbertForPreTraining.from_pretrained("albert-base-v2") input_ids = tf.constant([[0, 1, 2, 3, 4, 5]]) output = model(input_ids)[0] expected_shape = [1, 6, 30000] self.assertEqual(output.shape, expected_shape) expected_slice = tf.constant([[ [4.595668, 0.74462754, -1.818147], [4.5954347, 0.7454184, -1.8188258], [4.5954905, 0.7448235, -1.8182316], ]]) tf.debugging.assert_near(output[:, :3, :3], expected_slice, atol=1e-4)