OpenBuddy

Meta recently released a new generation of open source model series Llama3.1, including a 405B parameter version, whose performance is close to or even surpasses closed source models such as GPT-4 in some benchmarks. Llama3.1-8B-Instruct is an 8B parameter version in the series, supporting English, German, French, Italian, Portuguese, Spanish, Hindi and Thai, with context length up to 131072 tokens, and the knowledge deadline is updated to December 2023. To enhance Llama3.1-8B-Instr…

The OpenBuddy team released the latest achievement based on Meta Llama370B: OpenBuddy-Llama3-70B-v21.1-8k. This is their first attempt at Chinese cross-language training on the Llama370B base. Llama370B is a dense large model trained with a 15T training set. However, due to the lack of Chinese data in the training set, Meta's official Llama3 series dialogue model has weak cognitive and rational abilities in the Chinese field, and is prone to answering Chinese questions in English...