License needs clarification: use of PyMuPDF (aka fitz)

It is very obvious that this project makes use of PyMuPDF by [using the `fitz` package](https://github.com/deepseek-ai/DeepSeek-OCR/blob/e4ac34e1e59b891163fb9325480fbedec865e1f0/DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_pdf.py#L2) and [the addition of PyMuPDF in the requirements file](https://github.com/deepseek-ai/DeepSeek-OCR/blob/e4ac34e1e59b891163fb9325480fbedec865e1f0/requirements.txt#L3). PyMuPDF is licensed under the [AGPL-3.0 or a commercial license](https://pymupdf.readthedocs.io/en/latest/about.html#license-and-copyright). Thus, this project becomes a derivative work and would have to be licensed under the same terms (AGPL-3.0) due to the strong copyleft. I would like to ask you to clarify the license of this project (MIT).

The [preprint](https://doi.org/10.48550/arXiv.2510.18234) also mentions the use of _fitz_ - does this mean the model weights are also such a derivative work?