amad-iq/amad-vlm6-GGUF

amad-vlm6 — GGUF

GGUF builds of amad-iq/amad-vlm6, a TIES merge of two Arabic OCR fine-tunes of Qwen2.5-VL-7B-Instruct.

📄 Technical report · 🧠 Merged weights

The `mmproj` file is required alongside the model file — it carries the vision encoder. Download both into the same folder.

llama.cpp

llama-mtmd-cli -m amad-vlm6-q8_0.gguf --mmproj mmproj-amad-vlm6-f16.gguf \
  --image page.png -p "Extract the text in the image. Give me the final text, nothing else." \
  -n 4096 --temp 0 --repeat-penalty 1.05

LM Studio

Download a model file plus the mmproj into the same folder, load it, attach an image. Set context to at least 8192 and max output tokens to 4096 — the model may emit <think>…</think> first; keep only the text after the last </think>.

License

Apache 2.0.

Author: Murtadha · amad-iq