https://medium.com/@martin-thissen/llama-alpaca-chatgpt-on-your-local-computer-tutorial-17adda704c23

The LLaMA model is a fundamental language model that predicts the next word in a given sequence. – The LLaMA model is 13 times smaller than the GPT-3 model, but offers comparable or even better performance on most tests. – You can install the LLaMA model on your local computer using the dalai library. – The Alpaca model is an optimised version of the LLaMA model that is capable of following instructions, similar to the behaviour of ChatGPT. – The fine-tuning process for the Alpaca model cost less than $600, compared with the $5,000,000 needed to train the GPT-3 model in 2020. – The researchers used the unintended help of OpenAI to create sample instructions using the text-davinci-003 model. – You can install the Alpaca model on your local computer using the dalai library. – The dalai library also offers an API for integrating these models into custom applications. – The LLaMA and Alpaca models are available under non-commercial licence.