The paper discusses the Llama 2 project, developed by a team at Meta, including authors like Guillem Cucurull, Naman Goyal, and Thomas Scialom. It presents a collection of pretrained and fine-tuned large language models, emphasizing their performance in dialogue applications and their competitive edge over other open-source chat models. The Llama 2-Chat models are highlighted for their helpfulness and safety, based on extensive human evaluations.

свернуть