exp_classification里，所有的flag赋值用的都是大写字母的问题

[bsshor]

你好，如上图所示，在exp_classification里，所有的flag赋值用的都是大写字母。但是如下图在data_provider函数里，有一个判断标志是不是test的逻辑，用的是小写字母。我又比对了exp_anomaly_detection里关于flag的赋值，都是小写的test或train。请问为什么对exp_classification的flag赋值专门用大写呢？这个是bug吗？当我把exp_classification里，所有的flag赋值用的都更改为对应的小写字母后，再运行脚本，又会出现“Exception： No .ts files found using pattern： '*.ts”的报错 1711040776245

添加评论

[bsshor]

这是修改exp_classification里，所有的flag为小写后的报错： Traceback (most recent call last): File "/home/Time-Series-Library/run.py", line 128, in exp = Exp(args) # set experiments File "/home/Time-Series-Library/exp/exp_classification.py", line 18, in init super(Exp_Classification, self).init(args) File "/home/Time-Series-Library/exp/exp_basic.py", line 33, in init self.model = self._build_model().to(self.device) File "/home/Time-Series-Library/exp/exp_classification.py", line 22, in _build_model train_data, train_loader = self._get_data(flag='train') File "/home/Time-Series-Library/exp/exp_classification.py", line 35, in _get_data data_set, data_loader = data_provider(self.args, flag) File "/home/Time-Series-Library/data_provider/data_factory.py", line 57, in data_provider data_set = Data( File "/home/Time-Series-Library/data_provider/data_loader.py", line 619, in init self.all_df, self.labels_df = self.load_all(root_path, file_list=file_list, flag=flag) File "/home/Time-Series-Library/data_provider/data_loader.py", line 662, in load_all raise Exception("No .ts files found using pattern: '{}'".format(pattern)) Exception: No .ts files found using pattern: '*.ts'

[wuhaixu2016]

您好，这是因为数据集使用大写字母命名，所以其实传入的是数据集名称，不是flag 具体请见：https://github.com/thuml/Time-Series-Library/blob/main/data_provider/data_loader.py#L657-L658