Luth-LFM2-700M

Luth-LFM2-700M is a French fine-tuned version of LFM2-700M in collaboration with Liquid AI, trained on the Luth-SFT dataset. The model has improved its French capabilities in instruction following, math, and general knowledge. Additionally, its English capabilities have remained stable.

Our Evaluation, training and data scripts are available on GitHub, along with the Blog we wrote, to further detail our recipe.

Model Details

The model was trained using full fine-tuning on the Luth-SFT dataset with Axolotl. The resulting model was then merged back with LFM2-700M. This process successfully retained the model's English capabilities while improving its performance in French.

Benchmark Results

We used LightEval for evaluation, with custom tasks for the French benchmarks. The models were evaluated with a temperature=0.

French Benchmark Scores

Model	IFEval French	GPQA-Diamond French	MMLU French	Math500 French	Arc-Challenge French	Hellaswag French
Luth-LFM2-700M	50.22	27.92	44.72	38.40	36.70	48.25
LFM2-700M	41.96	20.81	43.70	32.40	36.27	41.51
Llama-3.2-1B	27.79	25.38	25.49	15.80	29.34	25.09
Qwen3-0.6B	44.86	26.90	27.13	29.20	31.57	25.10
Qwen2.5-0.5B-Instruct	22.00	25.89	35.04	12.00	28.23	51.45

English Benchmark Scores

Model	IFEval English	GPQA-Diamond English	MMLU English	Math500 English	Arc-Challenge English	Hellaswag English
Luth-LFM2-700M	63.40	29.29	50.39	38.40	38.91	54.05
LFM2-700M	65.06	30.81	50.65	32.00	38.65	52.54
Llama-3.2-1B	44.05	25.25	31.02	26.40	34.30	55.84
Qwen3-0.6B	57.18	29.29	36.79	43.40	33.70	42.92
Qwen2.5-0.5B-Instruct	29.70	29.29	43.80	32.00	32.17	49.56

Code Example

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("kurakurai/Luth-LFM2-700M")
model = AutoModelForCausalLM.from_pretrained("kurakurai/Luth-LFM2-700M")
messages = [
    {"role": "user", "content": "Quelle est la capitale de la France?"},
]
inputs = tokenizer.apply_chat_template(
    messages,
    add_generation_prompt=True,
    tokenize=True,
    return_dict=True,
    return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=100)
print(
    tokenizer.decode(
        outputs[0][inputs["input_ids"].shape[-1] :], skip_special_tokens=True
    )
)

Citation

@misc{luth2025kurakurai,
  title        = {Luth: Efficient French Specialization for Small Language Models and Cross-Lingual Transfer},
  author       = {Lasbordes, Maxence and Gad, Sinoué},
  year         = {2025},
  howpublished = {\url{https://arxiv.org/abs/2510.05846}},
  note         = {arXiv:2510.05846}
}