COLE Benchmark Introduces French Language Understanding Evaluation

The COLE benchmark evaluates French NLU on 23 tasks with 94 language models; open-weight models beat closed-weight ones, especially on zero‑shot QA. Read more: getnews.me/cole-benchmark-introduce... #frenchnlp #colebenchmark

LLM Prompting Beats Fine‑Tuning for French Medical Text Extraction

Prompting with GPT‑4.1 achieved the top macro‑F1 scores—61.53% for French biomedical NER and 15.02% for event extraction—in a few‑shot EvalLLM 2025 challenge (published 3 Oct 2025). Read more: getnews.me/llm-prompting-beats-fine... #frenchnlp #gpt41