Introducing Mistral 7B, the latest language model from the Mistral AI team. This model boasts an impressive 7.3B parameters and outperforms other models like Llama 2 13B and Llama 1 34B on various benchmarks. It excels in code and reasoning tasks, thanks to its use of Grouped-query attention (GQA) and Sliding Window Attention (SWA) for faster inference and handling longer sequences. Mistral 7B is available for download under the Apache 2.0 license and can be used without restrictions. It’s also easy to fine-tune for specific tasks, as demonstrated with the fine-tuned chat model that surpasses Llama 2 13B chat performance. Overall, Mistral 7B provides excellent performance and can be deployed on various platforms including cloud services.
https://mistral.ai/news/announcing-mistral-7b/