Do you also have the mmproj file?
I would assume the mmproj has also been finetuned for this model? I have the one from the alpha two checkpoint, but I don't think its the same.
Never mind, I ended up doing my own quants. Feel free to take the mmproj if you'd like.
https://huggingface.co/concedo/llama-joycaption-beta-one-hf-llava-mmproj-gguf
@concedo Awesome thanks a lot. How did you generate the mmproj files? We tried but it failed due to the following error likely because we didn’t use the legacy script:
{[[PROGRESS:mmproj extraction hf Q8_0]]}
INFO:hf-to-gguf:Loading model: llama-joycaption-beta-one-hf-llava
INFO:hf-to-gguf:Model architecture: SiglipVisionModel
ERROR:hf-to-gguf:Model SiglipVisionModel is not supported
job finished, status 1
job-done<0 llama-joycaption-beta-one-hf-llava noquant 1>
We had a discussion about this in https://huggingface.co/mradermacher/model_requests/discussions/937 and ended up almost nuking this model because of it but I then forgot about it. Now with an mmproj file actually existing this model is no longer useless. Thanks a lot for generating and providing them!
@nicoboss i explain the conversion process here: https://huggingface.co/mradermacher/model_requests/discussions/937#68260d236262358245dac1e7