mradermacher

/

model_requests

like 147

The PaddleOCRVLForConditionalGeneration architecture used by PaddlePaddle/PaddleOCR-VL is unfortunately not currently supported by llama.cpp and based on https://github.com/ggml-org/llama.cpp/pull/16701 probably never will be. There where efforts made to implement llama.cpp support for it but it was all for nothing as they came to the following conclusion:

The main issue is that PaddleOCR is not just one monolithic model like Qwen or Deepseek-OCR, but it's more like a pipeline of multiple models glued together. Therefore, I don't think we currently have the infrastructure to bring it into llama.cpp.