different result on training model and inference model on detection

[rubensanchezrivero]

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

• 系统环境/System Environment：Windows 11 x64
• 版本号/Version：Paddle：2.3.2 PaddleOCR：2.6.01 问题相关组件/Related components：
• 运行指令/Command Code：
• 完整报错/Complete Error Message： I fine tuned the detection model of ppocrv3, and the results of the trained model are just as expected, but after exporting it with
  python tools/export_model.py -c output/ch_PP-OCR_v3_det/config.yml -o Global.pretrained_model="./output/ch_PP-OCR_v3_det/best_accuracy" Global.save_inference_dir="./output/ch_PP-OCR_v3_det_inference/" the inference resulting model results are the same of the model before the training, could you please provide some insight about this issue please