clairedhx/mistral7b-labels2codes-lora

Model Card for mistral7b_labels2codes_lora

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3. It has been trained using TRL.

Mistral-7B Instruct — LoRA (ICD-10 labels → codes)

This LoRA adapter fine-tunes mistralai/Mistral-7B-Instruct-v0.3 to map French ICD-10 diagnostic labels (synonyms) to their corresponding codes (dot-less, up to 5 chars).

Quick start

from transformers import pipeline

question = "Libellé: Antécédent bronchite chronique"
generator = pipeline("text-generation", model="None", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])

Training procedure

It was trained via supervised fine-tuning (QLoRA) on an instruction dataset built from (label, code) pairs (instruct_labels2codes) derived from a curated ICD-10 synonyms table created from the webscrapinng of aideaucodage.fr .

Framework versions

• PEFT 0.17.0
• TRL: 0.21.0
• Transformers: 4.55.2
• Pytorch: 2.7.1
• Datasets: 3.6.0
• Tokenizers: 0.21.2

Citations

Cite TRL as:

@misc{vonwerra2022trl,
    title        = {{TRL: Transformer Reinforcement Learning}},
    author       = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
    year         = 2020,
    journal      = {GitHub repository},
    publisher    = {GitHub},
    howpublished = {\url{https://github.com/huggingface/trl}}
}