gpt2_ancient_translation_train.py

## gpt2 进行文言文翻译
import torch
from bert_seq2seq.utils import load_gpt
import sys

sys.path.append("/Users/xingzhaohu/Downloads/code/python/ml/ml_code/bert/bert_seq2seq")
import torch
from tqdm import tqdm
import torch.nn as nn
from torch.optim import Adam
import pandas as pd
import numpy as np
import os
import json
import time
import  glob
import bert_seq2seq
from torch.utils.data import Dataset, DataLoader
from bert_seq2seq.tokenizer import Tokenizer, load_chinese_base_vocab

vocab_path = "./state_dict/gpt2通用中文模型/vocab.txt"
model_path = "./state_dict/gpt2通用中文模型/pytorch_model.bin"
model_save_path = "./state_dict/gpt_ancient_trans_model.bin"
batch_size = 8
lr = 1e-5
word2idx = load_chinese_base_vocab(vocab_path)

def read_corpus():
    """
    读原始数据
    """
    src = []
    tgt = []
    data_path = glob.glob("./corpus/文言文翻译/*")
    for p in data_path:
        dir = p.split("/")[:-1]
        dir = "/".join(dir)
        # print(dir)
        name = p.split("/")[-1]
        if "翻译" in name:
            # 找到了一个翻译文件
            tgt_name = name
            src_name = name[:-2]
            with open(dir + "/" + src_name) as fs:
                lines = fs.readlines()
                for line in lines:
                    src.append(line.strip("\n").strip())

            with open(dir + "/" + tgt_name) as ft:
                lines = ft.readlines()
                for line in lines:
                    tgt.append(line.strip("\n").strip())

        else:
            pass

    return src, tgt

class SeqDataset(Dataset):
    """
    针对特定数据集，定义一个相关的取数据的方式
    """

    def __init__(self, sents_src, sents_tgt):
        ## 一般init函数是加载所有数据
        super(SeqDataset, self).__init__()
        # 读原始数据
        # self.sents_src, self.sents_tgt = read_corpus(poem_corpus_dir)
        self.sents_src = sents_src
        self.sents_tgt = sents_tgt

        self.idx2word = {k: v for v, k in word2idx.items()}
        self.tokenizer = Tokenizer(word2idx)

    def __getitem__(self, i):
        ## 得到单个数据
        # print(i)
        src = self.sents_src[i]
        tgt = self.sents_tgt[i]
        token_ids, _ = self.tokenizer.encode(src, tgt, max_length=256)
        output = {
            "token_ids": token_ids,
        }
        return output

    def __len__(self):
        return len(self.sents_src)


def collate_fn(batch):
    """
    动态padding， batch为一部分sample
    """

    def padding(indice, max_length, pad_idx=0):
        """
        pad 函数
        """
        pad_indice = [item + [pad_idx] * max(0, max_length - len(item)) for item in indice]
        return torch.tensor(pad_indice)

    token_ids = [data["token_ids"] for data in batch]
    max_length = max([len(t) for t in token_ids])

    token_ids_padded = padding(token_ids, max_length)
    target_ids_padded = token_ids_padded.clone()
    target_ids_padded[target_ids_padded == 0] = -100

    return token_ids_padded, target_ids_padded


class Trainer:
    def __init__(self):
        # 加载数据
        self.sents_src, self.sents_tgt = read_corpus()

        # 判断是否有可用GPU
        self.device = torch.device("cuda:3" if torch.cuda.is_available() else "cpu")
        print("device: " + str(self.device))
        # 定义模型
        self.gpt_model = load_gpt(word2idx)
        ## 加载预训练的模型参数～
        self.gpt_model.load_pretrain_params(model_path)
        # 将模型发送到计算设备(GPU或CPU)
        self.gpt_model.set_device(self.device)
        # 声明需要优化的参数
        self.optim_parameters = list(self.gpt_model.parameters())
        self.optimizer = torch.optim.Adam(self.optim_parameters, lr=lr, weight_decay=1e-3)
        # 声明自定义的数据加载器
        dataset = SeqDataset(self.sents_src, self.sents_tgt)
        self.dataloader = DataLoader(dataset, batch_size=batch_size, shuffle=True, collate_fn=collate_fn)

    def train(self, epoch):
        # 一个epoch的训练
        self.gpt_model.train()
        self.iteration(epoch, dataloader=self.dataloader, train=True)

    def save(self, save_path):
        """
        保存模型
        """
        self.gpt_model.save_all_params(save_path)
        print("{} saved!".format(save_path))

    def iteration(self, epoch, dataloader, train=True):
        total_loss = 0
        report_loss = 0
        start_time = time.time()  ## 得到当前时间
        step = 0
        # for token_ids, target_ids in tqdm(dataloader, position=0, leave=True):
        for token_ids, target_ids in dataloader:
            step += 1
            if step % 4000 == 0:
                self.gpt_model.eval()
                test_data = ["遂入颍川。", "会日暝，结陈相持。", "一言兴邦，斯近之矣。"]
                for text in test_data:
                    print(self.gpt_model.sample_generate(text, add_eos=True))
                self.gpt_model.train()
                print("report loss is " + str(report_loss))
                report_loss = 0.0

            # 因为传入了target标签，因此会计算loss并且返回
            loss, _ = self.gpt_model(token_ids,
                                                labels=target_ids,
                                                )
            # 反向传播
            if train:
                # 清空之前的梯度
                self.optimizer.zero_grad()
                # 反向传播, 获取新的梯度
                loss.backward()
                # 用获取的梯度更新模型参数
                self.optimizer.step()

            # 为计算当前epoch的平均loss
            total_loss += loss.item()
            report_loss += loss.item()

        end_time = time.time()
        spend_time = end_time - start_time
        # 打印训练信息
        print("epoch is " + str(epoch) + ". loss is " + str(total_loss) + ". spend time is " + str(spend_time))
        # 保存模型
        self.save(model_save_path)


if __name__ == '__main__':

    trainer = Trainer()
    train_epoches = 100
    for epoch in range(train_epoches):
        # 训练一个epoch
        trainer.train(epoch)