Qwen2.5-VL-72B-Instruct-mmproj-f16.gguf is broken

#4
by hp1337 - opened

I get gibberish with this file, and normal output with the ggml-org version of the file.

The mmproj in this repo is old now. It was built with git clone https://github.com/HimariO/llama.cpp.qwen2vl.git
cd llama.cpp.qwen2vl
git checkout qwen25-vl-20250404 . I will update . Where is the ggml-org version of this file . I looked here https://huggingface.co/ggml-org/Qwen2.5-VL-72B-Instruct-GGUF/tree/main no mmproj file.

My server is busy creating ggufs so I have uploaded new version of mmproj files taken from bartowski (reliable source) https://huggingface.co/bartowski/Qwen_Qwen2.5-VL-72B-Instruct-GGUF.

This comment has been hidden (marked as Spam)

Error: 400:

That is a html error code or?

Mungert changed discussion status to closed

Sign up or log in to comment