first commit

tests/models/mpnet/test_modeling_mpnet.py

@@ -0,0 +1,258 @@
+# Copyright 2020 The HuggingFace Inc. team, Microsoft Corporation.
+#
+# 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.
+
+
+import unittest
+
+from transformers import MPNetConfig, 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 (
+        MPNetForMaskedLM,
+        MPNetForMultipleChoice,
+        MPNetForQuestionAnswering,
+        MPNetForSequenceClassification,
+        MPNetForTokenClassification,
+        MPNetModel,
+    )
+
+
+class MPNetModelTester:
+    def __init__(
+        self,
+        parent,
+        batch_size=13,
+        seq_length=7,
+        is_training=True,
+        use_input_mask=True,
+        use_token_type_ids=False,
+        use_labels=True,
+        vocab_size=99,
+        hidden_size=64,
+        num_hidden_layers=2,
+        num_attention_heads=4,
+        intermediate_size=64,
+        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_token_type_ids = use_token_type_ids
+        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 get_config(self):
+        return MPNetConfig(
+            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,
+            initializer_range=self.initializer_range,
+        )
+
+    def create_and_check_mpnet_model(
+        self, config, input_ids, input_mask, sequence_labels, token_labels, choice_labels
+    ):
+        model = MPNetModel(config=config)
+        model.to(torch_device)
+        model.eval()
+        result = model(input_ids, input_mask)
+        result = model(input_ids)
+        self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))
+        self.parent.assertEqual(result.pooler_output.shape, (self.batch_size, self.hidden_size))
+
+    def create_and_check_mpnet_for_question_answering(
+        self, config, input_ids, input_mask, sequence_labels, token_labels, choice_labels
+    ):
+        model = MPNetForQuestionAnswering(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_mpnet_for_sequence_classification(
+        self, config, input_ids, input_mask, sequence_labels, token_labels, choice_labels
+    ):
+        config.num_labels = self.num_labels
+        model = MPNetForSequenceClassification(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_mpnet_for_multiple_choice(
+        self, config, input_ids, input_mask, sequence_labels, token_labels, choice_labels
+    ):
+        config.num_choices = self.num_choices
+        model = MPNetForMultipleChoice(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 create_and_check_mpnet_for_token_classification(
+        self, config, input_ids, input_mask, sequence_labels, token_labels, choice_labels
+    ):
+        config.num_labels = self.num_labels
+        model = MPNetForTokenClassification(config=config)
+        model.to(torch_device)
+        model.eval()
+        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 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
+
+
+@require_torch
+class MPNetModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
+    all_model_classes = (
+        (
+            MPNetForMaskedLM,
+            MPNetForMultipleChoice,
+            MPNetForQuestionAnswering,
+            MPNetForSequenceClassification,
+            MPNetForTokenClassification,
+            MPNetModel,
+        )
+        if is_torch_available()
+        else ()
+    )
+    pipeline_model_mapping = (
+        {
+            "feature-extraction": MPNetModel,
+            "fill-mask": MPNetForMaskedLM,
+            "text-classification": MPNetForSequenceClassification,
+            "token-classification": MPNetForTokenClassification,
+            "zero-shot": MPNetForSequenceClassification,
+        }
+        if is_torch_available()
+        else {}
+    )
+
+    test_resize_embeddings = True
+
+    def setUp(self):
+        self.model_tester = MPNetModelTester(self)
+        self.config_tester = ConfigTester(self, config_class=MPNetConfig, hidden_size=32)
+
+    def test_config(self):
+        self.config_tester.run_common_tests()
+
+    def test_mpnet_model(self):
+        config_and_inputs = self.model_tester.prepare_config_and_inputs()
+        self.model_tester.create_and_check_mpnet_model(*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_mpnet_for_sequence_classification(*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_mpnet_for_multiple_choice(*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_mpnet_for_token_classification(*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_mpnet_for_question_answering(*config_and_inputs)
+
+
+@require_torch
+class MPNetModelIntegrationTest(unittest.TestCase):
+    @slow
+    def test_inference_no_head(self):
+        model = MPNetModel.from_pretrained("microsoft/mpnet-base")
+        input_ids = torch.tensor([[0, 345, 232, 328, 740, 140, 1695, 69, 6078, 1588, 2]])
+        output = model(input_ids)[0]
+        expected_shape = torch.Size((1, 11, 768))
+        self.assertEqual(output.shape, expected_shape)
+        expected_slice = torch.tensor(
+            [[[-0.0550, 0.1943, -0.0740], [-0.0562, 0.2211, -0.0579], [-0.0437, 0.3337, -0.0641]]]
+        )
+        # compare the actual values for a slice.
+        torch.testing.assert_close(output[:, :3, :3], expected_slice, rtol=1e-4, atol=1e-4)

tests/models/mpnet/test_tokenization_mpnet.py

@@ -0,0 +1,32 @@
+# Copyright 2020 The HuggingFace Inc. team, Microsoft Corporation.
+#
+# 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.
+
+
+import unittest
+
+from transformers.models.mpnet.tokenization_mpnet import MPNetTokenizer
+from transformers.testing_utils import require_tokenizers
+
+from ...test_tokenization_common import TokenizerTesterMixin
+
+
+@require_tokenizers
+class MPNetTokenizerTest(TokenizerTesterMixin, unittest.TestCase):
+    from_pretrained_id = "microsoft/mpnet-base"
+    tokenizer_class = MPNetTokenizer
+
+    integration_expected_tokens = ['[UNK]', 'is', 'a', 'test', '[UNK]', '[UNK]', 'was', 'born', 'in', '92', '##00', '##0', ',', 'and', 'this', 'is', '[UNK]', '.', '生', '[UNK]', '的', '真', '[UNK]', '[UNK]', '[UNK]', '[UNK]', '[UNK]', '[UNK]', '[UNK]', '<s>', 'hi', '<s>', 'there', '[UNK]', 'following', 'string', 'should', 'be', 'properly', 'encoded', ':', '[UNK]', '.', '[UNK]', 'ir', '##d', 'and', '[UNK]', 'ir', '##d', '[UNK]', '[UNK]', 'how', 'are', 'you', 'doing']  # fmt: skip
+    integration_expected_token_ids = [104, 2007, 1041, 3235, 104, 104, 2005, 2145, 2003, 6231, 8893, 2696, 1014, 2002, 2027, 2007, 104, 1016, 1914, 104, 1920, 1925, 104, 104, 104, 104, 104, 104, 104, 0, 7636, 0, 2049, 104, 2210, 5168, 2327, 2026, 7923, 12363, 1028, 104, 1016, 104, 20872, 2098, 2002, 104, 20872, 2098, 104, 104, 2133, 2028, 2021, 2729]  # fmt: skip
+    expected_tokens_from_ids = ['[UNK]', 'is', 'a', 'test', '[UNK]', '[UNK]', 'was', 'born', 'in', '92', '##00', '##0', ',', 'and', 'this', 'is', '[UNK]', '.', '生', '[UNK]', '的', '真', '[UNK]', '[UNK]', '[UNK]', '[UNK]', '[UNK]', '[UNK]', '[UNK]', '<s>', 'hi', '<s>', 'there', '[UNK]', 'following', 'string', 'should', 'be', 'properly', 'encoded', ':', '[UNK]', '.', '[UNK]', 'ir', '##d', 'and', '[UNK]', 'ir', '##d', '[UNK]', '[UNK]', 'how', 'are', 'you', 'doing']  # fmt: skip
+    integration_expected_decoded_text = "[UNK] is a test [UNK] [UNK] was born in 92000, and this is [UNK]. 生 [UNK] 的 真 [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] <s> hi <s> there [UNK] following string should be properly encoded : [UNK]. [UNK] ird and [UNK] ird [UNK] [UNK] how are you doing"