Awesome multilingual OCR toolkits based on PaddlePaddle (practical ultra lightweight OCR system, support 80+ languages recognition, provide data annotation and synthesis tools, support training and deployment among server, mobile, embedded and IoT devices)
从日志中大概可以看出是训练的参数与模型不一致,但我是跟着教程一起做的,训练模型用的是https://paddleocr.bj.bcebos.com/PP-OCRv3/chinese/ch_PP-OCRv3_rec_train.tar。训练的配置文件是ch_PP-OCRv3_rec_distillation.yml训练参数在上面日志中有。请问是否改更换模型还是修改配置参数