args.py

import os
import torch
from transformers import BertTokenizer


class Config(object):

    """配置参数"""
    def __init__(self, data_path, bert_path):
        self.model_name = 'pytorch_bert_fine_tuing_fc_v1.0'
        self.save_path = data_path + '/saved_dict/' + self.model_name + '.ckpt' 
        self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
        self.class_list = ['0', '1']

        self.require_improvement = 1000
        self.num_classes = len(self.class_list)
        self.num_epochs = 4  
        self.batch_size = 128                       
        self.max_len = 64                                         
        self.learning_rate = 5e-4                             
        self.bert_path = bert_path
        print('Loading BERT tokenizer...')	
        self.tokenizer = BertTokenizer.from_pretrained(self.bert_path)
        self.hidden_size = 768