Llama-3.1-Tulu-3-70B-SFT is part of the Tülu3 model family, designed to provide comprehensive guidelines for modern post-training techniques. This model excels not only in chat tasks but also achieves state-of-the-art performance across various tasks including MATH, GSM8K, and IFEval. It is trained on publicly available, synthetic, and human-created datasets, primarily using English, and follows the Llama 3.1 community license agreement.