1M multi-lingual synthetic corporate emails to stress-test your RAG (huggingface.co)

A new dataset has been released by Sinabis GmbH containing over 1.25 million synthetic corporate emails, designed to stress-test retrieval-augmented generation (RAG) models. Spanning eight and a half years and generated by large language models, this dataset reflects the daily operations of a fictional energy company, featuring real attachments and detailed business cases. The dataset includes emails in multiple languages—English, German, French, and Italian—highlighting varied business topics including drilling operations, HR issues, and environmental matters. With 376,417 business cases documented, this exhaustive collection provides a rich resource for AI and machine learning practitioners. The significance of this dataset lies in its potential to enhance the performance and robustness of AI systems in processing and generating business communications. By utilizing structured and realistic examples of corporate interactions, researchers can better evaluate the capabilities of RAG models in understanding context, language detection, and the generation of appropriate responses. The dataset also reflects real-world email dynamics, such as the variety of topics and attachment types, thus offering valuable insights into the complexities of corporate communication which can be crucial for developing more sophisticated AI applications in fields like customer service and enterprise resource planning.