JY_L

Results 4 issues of JY_L

采用PPOCRLabelv2如何生成官方指定的如下数据集格式 zh_train_0.jpg [{"transcription": "汇丰晋信", "label": "other", "points": [[104, 114], [530, 114], [530, 175], [104, 175]], "id": 1, "linking": []}, {"transcription": "受理时间:", "label": "question", "points": [[126, 267], [266, 267], [266, 305],...

请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem - 版本号/Version:Paddle:2.6 - 运行指令/Command Code:python PPOCRLabel.py --lang ch --kie True - 标注结束后如何导出成kie需要的训练格式,采用那个脚本进行转化。自制数据集详细的操作流程

ser训练中前几轮次损失正常下降,后面就出现nanxxx,但是前几次的eval结果都是0.9以上 ![image](https://user-images.githubusercontent.com/67177370/202097432-dc497365-a3ba-4a0f-9190-a448b786c2f8.png)

![image](https://github.com/HUSTAI/uie_pytorch/assets/67177370/ffad3c90-0626-4804-8ed3-8f80c2db4610) ![image](https://github.com/HUSTAI/uie_pytorch/assets/67177370/f0671ad1-85e8-4943-a537-50a64152a0e9)