Daniel
> 在图像相关任务中，字符识别类型的训练通常需要较多的数据，如果数据实在不够，可以搜索下相关的印章生成脚本，一些参考资料可以参考rapidocr收集的[链接](https://rapidai.github.io/RapidOCRDocs/blog/2022/10/01/inference-summary/?h=%E5%8D%B0%E7%AB%A0#ocr_2)。 可以沟通下不？我这边是rapidOCR团队，微我下：364479, 开门密码：tr-ocr-seal
> @znsoftm > > * paddle2onnx don't support quanted model. > * if your model is contruct by paddlepaddle>2.0.0rc1，lstm is supported. Old version of paddlepaddle is not supported. > *...
The requirement is to transfer paddleocr mobile model V2.0 to onnx without input limitation.
目前转成ncnn有问题，只有部分模型工作, 参见 ： https://github.com/RapidOCR/RapidOCRNcnnCpp/tree/main/models
To be compatible with Bert-base/large, need a new parameter inputs_embeds to accept vectors as input.
I have done a research for binary file identification, originally it is based on bert-base/large. I would like to transfer it to ModernBert to determine the potential of our methods.
It is like a VLM, align binary data to language vector.
Thanks for your time.
Good idea. It is neccesary for researchers.
It suports webdav now.