The Ultimate Showdown: Mistral7b vs. Llama2–13b. Unveiling the Champion

Mistral7b is making the bold claim to be better than a model at least twice larger… Let’s check it out learning how to use it!

image created by the author

Meet the new kid on the block! Mistral 7B is a powerful language model with 7.3 billion parameters. It’s creating quite a buzz because it claims not only to beat Meta’s Llama 2–13b, but also to outshine other larger models. In this article, we’ll take a closer look at Mistral 7B, learn how to use it and compare the results with Llama2–13b-chat.

All we are going to do is only with FREE resources: we will see how to run the GGUF version (a quantized method that can run on any normal CPU) and the GPTQ version (aa quantized method that requires a GPU).

Mistral: a startup with a mission

Meet Mistral AI, an exciting new startup from Paris! Founded by brilliant minds who previously worked at tech giants like Google’s DeepMind and Meta, this company has made a splash in the industry. Their eye-catching Word Art logo and groundbreaking $118 million seed funding round have put them in the spotlight, making history in Europe.

But what’s their mission? It’s simple yet ambitious: to make AI truly beneficial…