Sao10K's picture
Create README.md
d8f1281 verified
Command to merge back (do within llama.cpp folder):
```
./gguf-split --merge /workspace/Franziska-Maxtral-8x22B-v1/Split-Franziska-Maxtral-8x22B-v1.q4_K_M-00001-of-00009.gguf /workspace/Franziska-Maxtral-8x22B-v1.q4_K_M.gguf
```
one gguf because its a testing model, fits in 2 A6000s fully at 16k context.
main info: https://huggingface.co/Sao10K/Franziska-Maxtral-8x22B-v1