def create_and_check_flaubert_model(
self,
config,
input_ids,
token_type_ids,
input_lengths,
sequence_labels,
token_labels,
is_impossible_labels,
choice_labels,
input_mask,
):
model = FlaubertModel(config=config)
model.to(torch_device)
model.eval()
outputs = model(input_ids, lengths=input_lengths, langs=token_type_ids)
outputs = model(input_ids, langs=token_type_ids)
outputs = model(input_ids)
sequence_output = outputs[0]
result = {
"sequence_output": sequence_output,
}
self.parent.assertListEqual(
list(result["sequence_output"].size()),
[self.batch_size, self.seq_length, self.hidden_size])
def __init__(self, config, pos_weight=None):
super(FlaubertForMultiLabelSequenceClassification, self).__init__(config)
self.num_labels = config.num_labels
self.pos_weight = pos_weight
self.transformer = FlaubertModel(config)
self.sequence_summary = SequenceSummary(config)
self.init_weights()
def create_and_check_flaubert_model(
self,
config,
input_ids,
token_type_ids,
input_lengths,
sequence_labels,
token_labels,
is_impossible_labels,
choice_labels,
input_mask,
):
model = FlaubertModel(config=config)
model.to(torch_device)
model.eval()
result = model(input_ids, lengths=input_lengths, langs=token_type_ids)
result = model(input_ids, langs=token_type_ids)
result = model(input_ids)
self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))