Luth-LFM2 are French fine-tuned versions of LFM2 models by Kurakura AI, trained on the Luth-SFT dataset. The models have improved their French capabilities in instruction following, math, and general knowledge. Additionally, their English capabilities have remained stable.

Find more details in the original model card: https://huggingface.co/kurakurai/Luth-LFM2-1.2B

🏃 How to run Luth-LFM2

Example usage with llama.cpp:

llama-cli -hf kurakurai/Luth-LFM2-1.2B-GGUF