Spaces:
Paused
Paused
# coding=utf-8 | |
# Copyright 2023 The HuggingFace Inc. and Baidu team. All rights reserved. | |
# | |
# Licensed under the Apache License, Version 2.0 (the "License"); | |
# you may not use this file except in compliance with the License. | |
# You may obtain a copy of the License at | |
# | |
# http://www.apache.org/licenses/LICENSE-2.0 | |
# | |
# Unless required by applicable law or agreed to in writing, software | |
# distributed under the License is distributed on an "AS IS" BASIS, | |
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | |
# See the License for the specific language governing permissions and | |
# limitations under the License. | |
""" Testing suite for the PyTorch ErnieM model. """ | |
import unittest | |
from transformers import ErnieMConfig, is_torch_available | |
from transformers.testing_utils import require_torch, slow, torch_device | |
from ...test_configuration_common import ConfigTester | |
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask | |
from ...test_pipeline_mixin import PipelineTesterMixin | |
if is_torch_available(): | |
import torch | |
from transformers import ( | |
ErnieMForInformationExtraction, | |
ErnieMForMultipleChoice, | |
ErnieMForQuestionAnswering, | |
ErnieMForSequenceClassification, | |
ErnieMForTokenClassification, | |
ErnieMModel, | |
) | |
from transformers.models.ernie_m.modeling_ernie_m import ERNIE_M_PRETRAINED_MODEL_ARCHIVE_LIST | |
class ErnieMModelTester: | |
def __init__( | |
self, | |
parent, | |
batch_size=13, | |
seq_length=7, | |
is_training=True, | |
use_input_mask=True, | |
use_labels=True, | |
vocab_size=99, | |
hidden_size=32, | |
num_hidden_layers=2, | |
num_attention_heads=4, | |
intermediate_size=37, | |
hidden_act="gelu", | |
hidden_dropout_prob=0.1, | |
attention_probs_dropout_prob=0.1, | |
max_position_embeddings=512, | |
type_vocab_size=16, | |
type_sequence_label_size=2, | |
initializer_range=0.02, | |
num_labels=3, | |
num_choices=4, | |
scope=None, | |
): | |
self.parent = parent | |
self.batch_size = batch_size | |
self.seq_length = seq_length | |
self.is_training = is_training | |
self.use_input_mask = use_input_mask | |
self.use_labels = use_labels | |
self.vocab_size = vocab_size | |
self.hidden_size = hidden_size | |
self.num_hidden_layers = num_hidden_layers | |
self.num_attention_heads = num_attention_heads | |
self.intermediate_size = intermediate_size | |
self.hidden_act = hidden_act | |
self.hidden_dropout_prob = hidden_dropout_prob | |
self.attention_probs_dropout_prob = attention_probs_dropout_prob | |
self.max_position_embeddings = max_position_embeddings | |
self.type_vocab_size = type_vocab_size | |
self.type_sequence_label_size = type_sequence_label_size | |
self.initializer_range = initializer_range | |
self.num_labels = num_labels | |
self.num_choices = num_choices | |
self.scope = scope | |
def prepare_config_and_inputs(self): | |
input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size) | |
input_mask = None | |
if self.use_input_mask: | |
input_mask = random_attention_mask([self.batch_size, self.seq_length]) | |
sequence_labels = None | |
token_labels = None | |
choice_labels = None | |
if self.use_labels: | |
sequence_labels = ids_tensor([self.batch_size], self.type_sequence_label_size) | |
token_labels = ids_tensor([self.batch_size, self.seq_length], self.num_labels) | |
choice_labels = ids_tensor([self.batch_size], self.num_choices) | |
config = self.get_config() | |
return config, input_ids, input_mask, sequence_labels, token_labels, choice_labels | |
def prepare_config_and_inputs_for_uiem(self): | |
input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size) | |
input_mask = None | |
if self.use_input_mask: | |
input_mask = random_attention_mask([self.batch_size, self.seq_length]) | |
config = self.get_config() | |
return config, input_ids, input_mask | |
def get_config(self): | |
return ErnieMConfig( | |
vocab_size=self.vocab_size, | |
hidden_size=self.hidden_size, | |
num_hidden_layers=self.num_hidden_layers, | |
num_attention_heads=self.num_attention_heads, | |
intermediate_size=self.intermediate_size, | |
hidden_act=self.hidden_act, | |
hidden_dropout_prob=self.hidden_dropout_prob, | |
attention_probs_dropout_prob=self.attention_probs_dropout_prob, | |
max_position_embeddings=self.max_position_embeddings, | |
type_vocab_size=self.type_vocab_size, | |
initializer_range=self.initializer_range, | |
) | |
def create_and_check_model(self, config, input_ids, input_mask, sequence_labels, token_labels, choice_labels): | |
model = ErnieMModel(config=config) | |
model.to(torch_device) | |
model.eval() | |
result = model(input_ids, return_dict=True) | |
self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size)) | |
def create_and_check_for_question_answering( | |
self, config, input_ids, input_mask, sequence_labels, token_labels, choice_labels | |
): | |
model = ErnieMForQuestionAnswering(config=config) | |
model.to(torch_device) | |
model.eval() | |
result = model( | |
input_ids, | |
attention_mask=input_mask, | |
start_positions=sequence_labels, | |
end_positions=sequence_labels, | |
) | |
self.parent.assertEqual(result.start_logits.shape, (self.batch_size, self.seq_length)) | |
self.parent.assertEqual(result.end_logits.shape, (self.batch_size, self.seq_length)) | |
def create_and_check_for_information_extraction( | |
self, config, input_ids, input_mask, sequence_labels, token_labels, choice_labels | |
): | |
model = ErnieMForInformationExtraction(config=config) | |
model.to(torch_device) | |
model.eval() | |
sequence_labels = torch.ones_like(input_ids, dtype=torch.float32) | |
result = model( | |
input_ids, | |
attention_mask=input_mask, | |
start_positions=sequence_labels, | |
end_positions=sequence_labels, | |
) | |
self.parent.assertEqual(result.start_logits.shape, (self.batch_size, self.seq_length)) | |
self.parent.assertEqual(result.end_logits.shape, (self.batch_size, self.seq_length)) | |
def create_and_check_for_sequence_classification( | |
self, config, input_ids, input_mask, sequence_labels, token_labels, choice_labels | |
): | |
config.num_labels = self.num_labels | |
model = ErnieMForSequenceClassification(config) | |
model.to(torch_device) | |
model.eval() | |
result = model(input_ids, attention_mask=input_mask, labels=sequence_labels) | |
self.parent.assertEqual(result.logits.shape, (self.batch_size, self.num_labels)) | |
def create_and_check_for_token_classification( | |
self, config, input_ids, input_mask, sequence_labels, token_labels, choice_labels | |
): | |
config.num_labels = self.num_labels | |
model = ErnieMForTokenClassification(config=config) | |
model.to(torch_device) | |
model.eval() | |
input_ids.to(torch_device) | |
input_mask.to(torch_device) | |
token_labels.to(torch_device) | |
result = model(input_ids, attention_mask=input_mask, labels=token_labels) | |
self.parent.assertEqual(result.logits.shape, (self.batch_size, self.seq_length, self.num_labels)) | |
def create_and_check_for_multiple_choice( | |
self, config, input_ids, input_mask, sequence_labels, token_labels, choice_labels | |
): | |
config.num_choices = self.num_choices | |
model = ErnieMForMultipleChoice(config=config) | |
model.to(torch_device) | |
model.eval() | |
multiple_choice_inputs_ids = input_ids.unsqueeze(1).expand(-1, self.num_choices, -1).contiguous() | |
multiple_choice_input_mask = input_mask.unsqueeze(1).expand(-1, self.num_choices, -1).contiguous() | |
result = model( | |
multiple_choice_inputs_ids, | |
attention_mask=multiple_choice_input_mask, | |
labels=choice_labels, | |
) | |
self.parent.assertEqual(result.logits.shape, (self.batch_size, self.num_choices)) | |
def prepare_config_and_inputs_for_common(self): | |
config_and_inputs = self.prepare_config_and_inputs() | |
( | |
config, | |
input_ids, | |
input_mask, | |
sequence_labels, | |
token_labels, | |
choice_labels, | |
) = config_and_inputs | |
inputs_dict = {"input_ids": input_ids, "attention_mask": input_mask} | |
return config, inputs_dict | |
class ErnieMModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase): | |
all_model_classes = ( | |
( | |
ErnieMModel, | |
ErnieMForMultipleChoice, | |
ErnieMForQuestionAnswering, | |
ErnieMForSequenceClassification, | |
ErnieMForTokenClassification, | |
) | |
if is_torch_available() | |
else () | |
) | |
all_generative_model_classes = () | |
pipeline_model_mapping = ( | |
{ | |
"feature-extraction": ErnieMModel, | |
"question-answering": ErnieMForQuestionAnswering, | |
"text-classification": ErnieMForSequenceClassification, | |
"token-classification": ErnieMForTokenClassification, | |
"zero-shot": ErnieMForSequenceClassification, | |
} | |
if is_torch_available() | |
else {} | |
) | |
test_torchscript = False | |
# TODO: Fix the failed tests when this model gets more usage | |
def is_pipeline_test_to_skip( | |
self, pipeline_test_casse_name, config_class, model_architecture, tokenizer_name, processor_name | |
): | |
if pipeline_test_casse_name == "QAPipelineTests": | |
return True | |
return False | |
def setUp(self): | |
self.model_tester = ErnieMModelTester(self) | |
self.config_tester = ConfigTester(self, config_class=ErnieMConfig, hidden_size=37) | |
def test_config(self): | |
self.config_tester.run_common_tests() | |
def test_model(self): | |
config_and_inputs = self.model_tester.prepare_config_and_inputs() | |
self.model_tester.create_and_check_model(*config_and_inputs) | |
def test_model_various_embeddings(self): | |
config_and_inputs = self.model_tester.prepare_config_and_inputs() | |
for type in ["absolute", "relative_key", "relative_key_query"]: | |
config_and_inputs[0].position_embedding_type = type | |
self.model_tester.create_and_check_model(*config_and_inputs) | |
def test_for_multiple_choice(self): | |
config_and_inputs = self.model_tester.prepare_config_and_inputs() | |
self.model_tester.create_and_check_for_multiple_choice(*config_and_inputs) | |
def test_for_question_answering(self): | |
config_and_inputs = self.model_tester.prepare_config_and_inputs() | |
self.model_tester.create_and_check_for_question_answering(*config_and_inputs) | |
def test_for_sequence_classification(self): | |
config_and_inputs = self.model_tester.prepare_config_and_inputs() | |
self.model_tester.create_and_check_for_sequence_classification(*config_and_inputs) | |
def test_for_information_extraction(self): | |
config_and_inputs = self.model_tester.prepare_config_and_inputs() | |
self.model_tester.create_and_check_for_information_extraction(*config_and_inputs) | |
def test_for_token_classification(self): | |
config_and_inputs = self.model_tester.prepare_config_and_inputs() | |
self.model_tester.create_and_check_for_token_classification(*config_and_inputs) | |
def test_model_from_pretrained(self): | |
for model_name in ERNIE_M_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: | |
model = ErnieMModel.from_pretrained(model_name) | |
self.assertIsNotNone(model) | |
class ErnieMModelIntegrationTest(unittest.TestCase): | |
def test_inference_model(self): | |
model = ErnieMModel.from_pretrained("susnato/ernie-m-base_pytorch") | |
model.eval() | |
input_ids = torch.tensor([[0, 1, 2, 3, 4, 5]]) | |
output = model(input_ids)[0] | |
# TODO Replace vocab size | |
hidden_size = 768 | |
expected_shape = torch.Size((1, 6, hidden_size)) | |
self.assertEqual(output.shape, expected_shape) | |
expected_slice = torch.tensor( | |
[[[-0.0012, 0.1245, -0.0214], [-0.0742, 0.0244, -0.0771], [-0.0333, 0.1164, -0.1554]]] | |
) | |
self.assertTrue(torch.allclose(output[:, :3, :3], expected_slice, atol=1e-3)) | |