1. transformers包加载预训练好的Bert模型

# 1. 导入包
import torch
from transformers import BertTokenizer

# 2. 所需要的预训练好的model
model_name = 'bert-base-uncased'

# 3. 通过词典导入分词器
tokenizer = BertTokenizer.from_pretrained(model_name)
sentence = "A very clean and well decorated empty bathroom."

2. 得到句子Embedding

（1）encode()方法：仅返回input_ids

def encode(
        self,
        text: Union[TextInput, PreTokenizedInput, EncodedInput],
        text_pair: Optional[Union[TextInput, PreTokenizedInput, EncodedInput]] = None,
        add_special_tokens: bool = True,
        padding: Union[bool, str, PaddingStrategy] = False,
        truncation: Union[bool, str, TruncationStrategy] = False,
        max_length: Optional[int] = None,
        stride: int = 0,
        return_tensors: Optional[Union[str, TensorType]] = None,
        **kwargs
    )

（2）encode_plus()方法:返回所有的编码信息

 def encode_plus(
        self,
        text: Union[TextInput, PreTokenizedInput, EncodedInput],
        text_pair: Optional[Union[TextInput, PreTokenizedInput, EncodedInput]] = None,
        add_special_tokens: bool = True,
        padding: Union[bool, str, PaddingStrategy] = False,
        truncation: Union[bool, str, TruncationStrategy] = False,
        max_length: Optional[int] = None,
        stride: int = 0,
        is_split_into_words: bool = False,
        pad_to_multiple_of: Optional[int] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        return_token_type_ids: Optional[bool] = None,
        return_attention_mask: Optional[bool] = None,
        return_overflowing_tokens: bool = False,
        return_special_tokens_mask: bool = False,
        return_offsets_mapping: bool = False,
        return_length: bool = False,
        verbose: bool = True,
        **kwargs
    )

一般返回参数如下：

1. input_ids：表示单词在词典中的编码id
2. token_type_ids：区分两个句子的编码id（上句全为0，下句全为1）
3. attention_mask：指定对哪些单词id进行self-Attention操作

• 举例如下：
• （1）encode()方法

# 4. 得到Embedding
# （1）encode()方法
print(tokenizer.encode(sentence))# encode仅返回input_ids
# [101, 7592, 1010, 2026, 2365, 2003, 5870, 1012, 102]

• （2）encode_plus()方法

# （2）encode_plus()方法
print(tokenizer.encode_plus(sentence))# encode_plus返回所有的编码信息

• （3）encode_plus()里面参数自定义

# （3）encode_plus()里面参数自定义
print(tokenizer.encode_plus(sentence, max_length=15, padding='max_length', return_attention_mask=True, return_token_type_ids=True, truncation=True))

3. Eg：以上代码整理，可跑

import torch
from transformers import BertTokenizer

model_name = 'bert-base-uncased'

# a.通过词典导入分词器
tokenizer = BertTokenizer.from_pretrained(model_name)
sentence = "A very clean and well decorated empty bathroom."

print(tokenizer.encode(sentence))# encode仅返回input_ids
# [101, 7592, 1010, 2026, 2365, 2003, 5870, 1012, 102]
print(tokenizer.encode_plus(sentence))# encode_plus返回所有的编码信息
print(tokenizer.encode_plus(sentence, max_length=15, padding='max_length', return_attention_mask=True, return_token_type_ids=True, truncation=True))
# {'input_ids': [101, 1037, 2200, 4550, 1998, 2092, 7429, 4064, 5723, 1012, 102], 'token_type_ids': [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], 'attention_mask': [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]}
# {'input_ids': [101, 1037, 2200, 4550, 1998, 2092, 7429, 4064, 5723, 1012, 102, 0, 0, 0, 0], 'token_type_ids': [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, 0, 0, 0, 0]}