MedInjection-FR/ALL

MedInjection-FR是一个大规模的法语生物医学指令数据集，旨在研究数据来源（原生、合成或翻译监督）对大型语言模型（LLM）指令微调的影响。它为适应法语医疗推理、问答和指令遵循任务的大型语言模型提供了一个统一基准。MedInjection-FR is a large-scale French biomedical instruction dataset designed to study how data provenance — native, synthetic, or translated supervision — impacts instruction tuning of large language models (LLMs). It provides a unified benchmark for adapting LLMs to French medical reasoning, question answering, and instruction-following tasks.