使用无标注的数据训练Bert

爱被打了一巴掌 2023-10-09 21:26 34阅读 0赞

#### 文章目录 ####

*  1、准备用于训练的数据集
 *  2、处理数据集
 *  3、克隆代码
 *  4、运行代码
 *  5、将ckpt模型转为bin模型使其可在pytorch中运用

Bert官方仓库：[https://github.com/google-research/bert][https_github.com_google-research_bert]

## 1、准备用于训练的数据集 ##

此处准备的是BBC news的数据集，下载链接：[https://www.kaggle.com/datasets/gpreda/bbc-news][https_www.kaggle.com_datasets_gpreda_bbc-news]  
原数据集格式（.csv）：  
![在这里插入图片描述][98f05a0678a04916b99f78444735f772.png]

## 2、处理数据集 ##

训练Bert时需要预处理数据，将数据处理成[https://github.com/google-research/bert/blob/master/sample\_text.txt][https_github.com_google-research_bert_blob_master_sample_text.txt]中所示格式，如下所示：  
![在这里插入图片描述][ab0cfa011fbc4e2f9a8b00f260b252b5.png]  
数据预处理代码参考：

import pandas as pd
    
    # 读取BBC-news数据集
    df = pd.read_csv("../../bbc_news.csv")
    # print(df['title'])
    l1 = []
    l2 = []
    cnt = 0
    for line in df['title']:
        l1.append(line)
    
    for line in df['description']:
        l2.append(line)
    # cnt=0
    f = open("test1.txt", 'w+', encoding='utf8')
    for i in range(len(l1)):
        s = l1[i] + " " + l2[i] + '\n'
        f.write(s)
        # cnt+=1
        # if cnt>10: break
    f.close()
    # print(l1)

处理完后的BBC news数据集格式如下所示：  
![在这里插入图片描述][b595597c4f5e40a49f0e326b920520d5.png]

## 3、克隆代码 ##

使用git克隆仓库代码  
http:

git clone https://github.com/google-research/bert.git

或ssh:

git clone git@github.com:google-research/bert.git

## 4、运行代码 ##

先下载Bert模型：[BERT-Base, Uncased][BERT-Base_ Uncased]  
该文件中有以下文件：  
![在这里插入图片描述][722625fe6a1e41aa95632bd85554168f.png]  
运行代码：  
在Teminal中运行：

python create_pretraining_data.py \
      --input_file=./sample_text.txt（数据集地址） \
      --output_file=/tmp/tf_examples.tfrecord（处理后数据集保存的位置） \
      --vocab_file=$BERT_BASE_DIR/vocab.txt（vocab.txt文件位置） \
      --do_lower_case=True \
      --max_seq_length=128 \
      --max_predictions_per_seq=20 \
      --masked_lm_prob=0.15 \
      --random_seed=12345 \
      --dupe_factor=5

训练模型：

python run_pretraining.py \
      --input_file=/tmp/tf_examples.tfrecord（处理后数据集保存的位置） \
      --output_dir=/tmp/pretraining_output（训练后模型保存位置） \
      --do_train=True \
      --do_eval=True \
      --bert_config_file=$BERT_BASE_DIR/bert_config.json（bert_config.json文件位置） \
      --init_checkpoint=$BERT_BASE_DIR/bert_model.ckpt（如果要从头开始的预训练，则去掉这行） \
      --train_batch_size=32 \
      --max_seq_length=128 \
      --max_predictions_per_seq=20 \
      --num_train_steps=20 \
      --num_warmup_steps=10 \
      --learning_rate=2e-5

训练完成后模型输出示例：

***** Eval results *****
      global_step = 20
      loss = 0.0979674
      masked_lm_accuracy = 0.985479
      masked_lm_loss = 0.0979328
      next_sentence_accuracy = 1.0
      next_sentence_loss = 3.45724e-05

要注意应该能够在至少具有 12GB RAM 的 GPU 上运行，不然会报错显存不足。  
[使用未标注数据训练BERT][BERT]

## 5、将ckpt模型转为bin模型使其可在pytorch中运用 ##

上一步训练好后准备好训练出来的model.ckpt-20.index文件和Bert模型中的bert\_config.json文件

创建python文件convert\_bert\_original\_tf\_checkpoint\_to\_pytorch.py：

# coding=utf-8
    # Copyright 2018 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.
    """Convert BERT checkpoint."""
    
    
    import argparse
    
    import torch
    
    from transformers import BertConfig, BertForPreTraining, load_tf_weights_in_bert
    from transformers.utils import logging
    
    
    logging.set_verbosity_info()
    
    
    def convert_tf_checkpoint_to_pytorch(tf_checkpoint_path, bert_config_file, pytorch_dump_path):
        # Initialise PyTorch model
        config = BertConfig.from_json_file(bert_config_file)
        print("Building PyTorch model from configuration: {}".format(str(config)))
        model = BertForPreTraining(config)
    
        # Load weights from tf checkpoint
        load_tf_weights_in_bert(model, config, tf_checkpoint_path)
    
        # Save pytorch-model
        print("Save PyTorch model to {}".format(pytorch_dump_path))
        torch.save(model.state_dict(), pytorch_dump_path)
    
    
    if __name__ == "__main__":
        parser = argparse.ArgumentParser()
        # Required parameters
        parser.add_argument(
            "--tf_checkpoint_path", default=None, type=str, required=True, help="Path to the TensorFlow checkpoint path."
        )
        parser.add_argument(
            "--bert_config_file",
            default=None,
            type=str,
            required=True,
            help="The config json file corresponding to the pre-trained BERT model. \n"
            "This specifies the model architecture.",
        )
        parser.add_argument(
            "--pytorch_dump_path", default=None, type=str, required=True, help="Path to the output PyTorch model."
        )
        args = parser.parse_args()
        convert_tf_checkpoint_to_pytorch(args.tf_checkpoint_path, args.bert_config_file, args.pytorch_dump_path)

在Terminal中运行以下命令：

python convert_bert_original_tf_checkpoint_to_pytorch.py \
    --tf_checkpoint_path Models/chinese_L-12_H-768_A-12/bert_model.ckpt.index（.ckpt.index文件位置） \
    --bert_config_file Models/chinese_L-12_H-768_A-12/bert_config.json（bert_config.json文件位置）  \
    --pytorch_dump_path  Models/chinese_L-12_H-768_A-12/pytorch_model.bin（输出的.bin模型文件位置）

以上命令最好在一行中运行：

python convert_bert_original_tf_checkpoint_to_pytorch.py --tf_checkpoint_path bert_model.ckpt.index --bert_config_file bert_config.json  --pytorch_dump_path  pytorch_model.bin

然后就可以得到bin文件了  
![在这里插入图片描述][aab8bfda15824b81922d58503978132d.png]

[【BERT for Tensorflow】本地ckpt文件的BERT使用][BERT for Tensorflow_ckpt_BERT]

[https_github.com_google-research_bert]: https://github.com/google-research/bert
[https_www.kaggle.com_datasets_gpreda_bbc-news]: https://www.kaggle.com/datasets/gpreda/bbc-news
[98f05a0678a04916b99f78444735f772.png]: https://img-blog.csdnimg.cn/98f05a0678a04916b99f78444735f772.png
[https_github.com_google-research_bert_blob_master_sample_text.txt]: https://github.com/google-research/bert/blob/master/sample_text.txt
[ab0cfa011fbc4e2f9a8b00f260b252b5.png]: https://img-blog.csdnimg.cn/ab0cfa011fbc4e2f9a8b00f260b252b5.png
[b595597c4f5e40a49f0e326b920520d5.png]: https://img-blog.csdnimg.cn/b595597c4f5e40a49f0e326b920520d5.png
[BERT-Base_ Uncased]: https://storage.googleapis.com/bert_models/2018_10_18/uncased_L-12_H-768_A-12.zip
[722625fe6a1e41aa95632bd85554168f.png]: https://img-blog.csdnimg.cn/722625fe6a1e41aa95632bd85554168f.png
[BERT]: https://blog.csdn.net/weixin_41089007/article/details/105596476
[aab8bfda15824b81922d58503978132d.png]: https://img-blog.csdnimg.cn/aab8bfda15824b81922d58503978132d.png
[BERT for Tensorflow_ckpt_BERT]: https://blog.csdn.net/qq_44574333/article/details/109631559?spm=1001.2014.3001.5506