EngineX-Cambricon/enginex-mlu370-any2any
9
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

init

Browse Source
This commit is contained in:
luopingyi
2025-10-09 16:47:16 +08:00
parent c8feb4deb5
commit e27e3f16bb
5248 changed files with 1778505 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

0
transformers/tests/models/deberta/__init__.py Normal file
View File

308
transformers/tests/models/deberta/test_modeling_deberta.py Normal file
View File

@@ -0,0 +1,308 @@
# Copyright 2018 Microsoft Authors and the HuggingFace Inc. team.
#
# 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 DebertaConfig, is_torch_available
from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
DebertaForMaskedLM,
DebertaForQuestionAnswering,
DebertaForSequenceClassification,
DebertaForTokenClassification,
DebertaModel,
)
class DebertaModelTester:
def __init__(
self,
parent,
batch_size=13,
seq_length=7,
is_training=True,
use_input_mask=True,
use_token_type_ids=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,
relative_attention=False,
position_biased_input=True,
pos_att_type="None",
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.relative_attention = relative_attention
self.position_biased_input = position_biased_input
self.pos_att_type = pos_att_type
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 = ids_tensor([self.batch_size, self.seq_length], vocab_size=2)
token_type_ids = None
if self.use_token_type_ids:
token_type_ids = ids_tensor([self.batch_size, self.seq_length], self.type_vocab_size)
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, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def get_config(self):
return DebertaConfig(
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,
relative_attention=self.relative_attention,
position_biased_input=self.position_biased_input,
pos_att_type=self.pos_att_type,
)
def get_pipeline_config(self):
config = self.get_config()
config.vocab_size = 300
return config
def check_loss_output(self, result):
self.parent.assertListEqual(list(result.loss.size()), [])
def create_and_check_deberta_model(
self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
):
model = DebertaModel(config=config)
model.to(torch_device)
model.eval()
sequence_output = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids)[0]
sequence_output = model(input_ids, token_type_ids=token_type_ids)[0]
sequence_output = model(input_ids)[0]
self.parent.assertListEqual(list(sequence_output.size()), [self.batch_size, self.seq_length, self.hidden_size])
def create_and_check_deberta_for_masked_lm(
self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
):
model = DebertaForMaskedLM(config=config)
model.to(torch_device)
model.eval()
result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=token_labels)
self.parent.assertEqual(result.logits.shape, (self.batch_size, self.seq_length, self.vocab_size))
def create_and_check_deberta_for_sequence_classification(
self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
):
config.num_labels = self.num_labels
model = DebertaForSequenceClassification(config)
model.to(torch_device)
model.eval()
result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=sequence_labels)
self.parent.assertListEqual(list(result.logits.size()), [self.batch_size, self.num_labels])
self.check_loss_output(result)
def create_and_check_deberta_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 = DebertaForTokenClassification(config=config)
model.to(torch_device)
model.eval()
result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=token_labels)
self.parent.assertEqual(result.logits.shape, (self.batch_size, self.seq_length, self.num_labels))
def create_and_check_deberta_for_question_answering(
self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
):
model = DebertaForQuestionAnswering(config=config)
model.to(torch_device)
model.eval()
result = model(
input_ids,
attention_mask=input_mask,
token_type_ids=token_type_ids,
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 prepare_config_and_inputs_for_common(self):
config_and_inputs = self.prepare_config_and_inputs()
(
config,
input_ids,
token_type_ids,
input_mask,
sequence_labels,
token_labels,
choice_labels,
) = config_and_inputs
inputs_dict = {"input_ids": input_ids, "token_type_ids": token_type_ids, "attention_mask": input_mask}
return config, inputs_dict
@require_torch
class DebertaModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
all_model_classes = (
(
DebertaModel,
DebertaForMaskedLM,
DebertaForSequenceClassification,
DebertaForTokenClassification,
DebertaForQuestionAnswering,
)
if is_torch_available()
else ()
)
pipeline_model_mapping = (
{
"feature-extraction": DebertaModel,
"fill-mask": DebertaForMaskedLM,
"question-answering": DebertaForQuestionAnswering,
"text-classification": DebertaForSequenceClassification,
"token-classification": DebertaForTokenClassification,
"zero-shot": DebertaForSequenceClassification,
}
if is_torch_available()
else {}
)
fx_compatible = True
test_torchscript = False
test_pruning = False
test_head_masking = False
is_encoder_decoder = False
def setUp(self):
self.model_tester = DebertaModelTester(self)
self.config_tester = ConfigTester(self, config_class=DebertaConfig, hidden_size=37)
def test_config(self):
self.config_tester.run_common_tests()
def test_deberta_model(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_deberta_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_deberta_for_sequence_classification(*config_and_inputs)
def test_for_masked_lm(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_deberta_for_masked_lm(*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_deberta_for_question_answering(*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_deberta_for_token_classification(*config_and_inputs)
@slow
def test_model_from_pretrained(self):
model_name = "microsoft/deberta-base"
model = DebertaModel.from_pretrained(model_name)
self.assertIsNotNone(model)
@unittest.skip("This test was broken by the refactor in #22105, TODO @ArthurZucker")
def test_torch_fx_output_loss(self):
pass
@unittest.skip("This test was broken by the refactor in #22105, TODO @ArthurZucker")
def test_torch_fx(self):
pass
@require_torch
@require_sentencepiece
@require_tokenizers
class DebertaModelIntegrationTest(unittest.TestCase):
@unittest.skip(reason="Model not available yet")
def test_inference_masked_lm(self):
pass
@slow
def test_inference_no_head(self):
model = DebertaModel.from_pretrained("microsoft/deberta-base")
input_ids = torch.tensor([[0, 31414, 232, 328, 740, 1140, 12695, 69, 46078, 1588, 2]])
attention_mask = torch.tensor([[0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]])
with torch.no_grad():
output = model(input_ids, attention_mask=attention_mask)[0]
# compare the actual values for a slice.
expected_slice = torch.tensor(
[[[-0.5986, -0.8055, -0.8462], [1.4484, -0.9348, -0.8059], [0.3123, 0.0032, -1.4131]]]
)
torch.testing.assert_close(output[:, 1:4, 1:4], expected_slice, rtol=1e-4, atol=1e-4)

170
transformers/tests/models/deberta/test_tokenization_deberta.py Normal file
View File

@@ -0,0 +1,170 @@
# Copyright 2019 Hugging Face inc.
#
# 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 json
import os
import unittest
from transformers import DebertaTokenizer, DebertaTokenizerFast
from transformers.models.deberta.tokenization_deberta import VOCAB_FILES_NAMES
from transformers.testing_utils import slow
from ...test_tokenization_common import TokenizerTesterMixin
class DebertaTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
from_pretrained_id = "microsoft/deberta-base"
tokenizer_class = DebertaTokenizer
test_rust_tokenizer = True
rust_tokenizer_class = DebertaTokenizerFast
@classmethod
def setUpClass(cls):
super().setUpClass()
# Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
vocab = [
"l",
"o",
"w",
"e",
"r",
"s",
"t",
"i",
"d",
"n",
"\u0120",
"\u0120l",
"\u0120n",
"\u0120lo",
"\u0120low",
"er",
"\u0120lowest",
"\u0120newer",
"\u0120wider",
"[UNK]",
]
vocab_tokens = dict(zip(vocab, range(len(vocab))))
merges = ["#version: 0.2", "\u0120 l", "\u0120l o", "\u0120lo w", "e r", ""]
cls.special_tokens_map = {"unk_token": "[UNK]"}
cls.vocab_file = os.path.join(cls.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
cls.merges_file = os.path.join(cls.tmpdirname, VOCAB_FILES_NAMES["merges_file"])
with open(cls.vocab_file, "w", encoding="utf-8") as fp:
fp.write(json.dumps(vocab_tokens) + "\n")
with open(cls.merges_file, "w", encoding="utf-8") as fp:
fp.write("\n".join(merges))
@classmethod
def get_tokenizer(cls, pretrained_name=None, **kwargs):
kwargs.update(cls.special_tokens_map)
pretrained_name = pretrained_name or cls.tmpdirname
return cls.tokenizer_class.from_pretrained(pretrained_name, **kwargs)
def get_input_output_texts(self, tokenizer):
input_text = "lower newer"
output_text = "lower newer"
return input_text, output_text
def test_full_tokenizer(self):
tokenizer = self.get_tokenizer()
text = "lower newer"
bpe_tokens = ["l", "o", "w", "er", "\u0120", "n", "e", "w", "er"]
tokens = tokenizer.tokenize(text)
self.assertListEqual(tokens, bpe_tokens)
input_tokens = tokens + [tokenizer.unk_token]
input_bpe_tokens = [0, 1, 2, 15, 10, 9, 3, 2, 15, 19]
self.assertListEqual(tokenizer.convert_tokens_to_ids(input_tokens), input_bpe_tokens)
def test_token_type_ids(self):
tokenizer = self.get_tokenizer()
tokd = tokenizer("Hello", "World")
expected_token_type_ids = [0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1]
self.assertListEqual(tokd["token_type_ids"], expected_token_type_ids)
@slow
def test_sequence_builders(self):
tokenizer = self.tokenizer_class.from_pretrained("microsoft/deberta-base")
text = tokenizer.encode("sequence builders", add_special_tokens=False)
text_2 = tokenizer.encode("multi-sequence build", add_special_tokens=False)
encoded_text_from_decode = tokenizer.encode(
"sequence builders", add_special_tokens=True, add_prefix_space=False
)
encoded_pair_from_decode = tokenizer.encode(
"sequence builders", "multi-sequence build", add_special_tokens=True, add_prefix_space=False
)
encoded_sentence = tokenizer.build_inputs_with_special_tokens(text)
encoded_pair = tokenizer.build_inputs_with_special_tokens(text, text_2)
assert encoded_sentence == encoded_text_from_decode
assert encoded_pair == encoded_pair_from_decode
@slow
def test_tokenizer_integration(self):
tokenizer_classes = [self.tokenizer_class]
if self.test_rust_tokenizer:
tokenizer_classes.append(self.rust_tokenizer_class)
for tokenizer_class in tokenizer_classes:
tokenizer = tokenizer_class.from_pretrained("microsoft/deberta-base")
sequences = [
"ALBERT: A Lite BERT for Self-supervised Learning of Language Representations",
"ALBERT incorporates two parameter reduction techniques",
"The first one is a factorized embedding parameterization. By decomposing the large vocabulary"
" embedding matrix into two small matrices, we separate the size of the hidden layers from the size of"
" vocabulary embedding.",
]
encoding = tokenizer(sequences, padding=True)
decoded_sequences = [tokenizer.decode(seq, skip_special_tokens=True) for seq in encoding["input_ids"]]
# fmt: off
expected_encoding = {
'input_ids': [
[1, 2118, 11126, 565, 35, 83, 25191, 163, 18854, 13, 12156, 12, 16101, 25376, 13807, 9, 22205, 27893, 1635, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[1, 2118, 11126, 565, 24536, 80, 43797, 4878, 7373, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[1, 133, 78, 65, 16, 10, 3724, 1538, 33183, 11303, 43797, 1938, 4, 870, 24165, 29105, 5, 739, 32644, 33183, 11303, 36173, 88, 80, 650, 7821, 45940, 6, 52, 2559, 5, 1836, 9, 5, 7397, 13171, 31, 5, 1836, 9, 32644, 33183, 11303, 4, 2]
],
'token_type_ids': [
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
],
'attention_mask': [
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
]
}
# fmt: on
expected_decoded_sequence = [
"ALBERT: A Lite BERT for Self-supervised Learning of Language Representations",
"ALBERT incorporates two parameter reduction techniques",
"The first one is a factorized embedding parameterization. By decomposing the large vocabulary"
" embedding matrix into two small matrices, we separate the size of the hidden layers from the size of"
" vocabulary embedding.",
]
self.assertDictEqual(encoding.data, expected_encoding)
for expected, decoded in zip(expected_decoded_sequence, decoded_sequences):
self.assertEqual(expected, decoded)