Researchers have created a benchmark for classifying French news articles from different publishers. The FrenchNews-7 benchmark combines a large corpus of news articles with a taxonomy and a fine-tuned classifier called CamemBERT. The study evaluates the performance of various classifiers, including zero-shot language models, on this benchmark. The results show that CamemBERT outperforms other models in certain categories, but also highlights inconsistencies in classification across different publishers. --- Why it matters: This research matters to engineers and researchers working on natural language processing tasks because it provides a new benchmark for evaluating the performance of classifiers on French news articles. Understanding how well these classifiers can generalize to unseen data is crucial for developing reliable AI systems that can handle real-world tasks. Source: https://arxiv.org/abs/2608.18097