- https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407
- Installer Python 3.12 : winget install Python.Python.3.12
- Installer xformers pour CUDA 1.24 : pip install -U xformers --index-url https://download.pytorch.org/whl/cu124
- Installer triton pour windows : pip install https://huggingface.co/madbuda/triton-windows-builds/resolve/main/triton-3.0.0-cp312-cp312-win_amd64.whl
- Installer pytorch pour CUDA 1.24 : pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124
- Installer huggingface_hub : pip3 install huggingface_hub
- Se connecter sur hugginggfacehub: huggingface-cli login
- Passer Python en UTF-8 : set PYTHONUTF8=1 && set PYTHONIOENCODING=utf-8
- Lancer mistral-chat ./mistral_models/Nemo-Instruct --instruct --max_tokens 256 --temperature 0.35