Pre-training and fine-tuning datasets for PP-OCRv5

[awkrail]

Thank you for sharing your excellent work!
I’m trying to reproduce the PP-OCRv5_mobile_det model, starting from pre-training all the way to fine-tuning.

According to the configuration YAML file, it looks like the fine-tuning is done using the ICDAR2015 dataset.
However, I couldn’t find any information about the dataset used for pre-training.
Could you let me know which dataset was used to pre-train the model?

Specifically, I’d like to understand how the following pretrained model was constructed:
PPLCNetV3_x0_75_ocr_det.pdparams

Thanks in advance!