clef-flash (GGUF)
GGUF weights for clef-flash, exported from a local Ollama pull for use with llama.cpp / Ollama / LM Studio and other GGUF-compatible runtimes.
Files
| File | Description | Size |
|---|---|---|
clef-flash-Q8_0.gguf |
Main model weights, Q8_0 quantization | ~10 GB |
clef-flash-vision.gguf |
Vision projector (multimodal) | ~918 MB |
Usage
See RUN.md for full Docker instructions (Ollama and llama.cpp server-cuda, incl. A100/CUDA 12 serving).
Quick start with Ollama:
ollama create clef-flash -f Modelfile
ollama run clef-flash
Or run directly with llama.cpp:
llama-cli -m clef-flash-Q8_0.gguf --mmproj clef-flash-vision.gguf -c 16384
License
Apache License 2.0 (see LICENSE).