ChatLLaMA

Meta recently unveiled LLaMA, an array of foundational large language models with parameters ranging from 7 to 65 billion. LLaMA generates a lot of enthusiasm due to it being smaller than GPT-3 yet displaying superior performance. For instance, LLaMA's 13B model outperforms GPT-3 despite being ten times smaller. This new set of basic models provides the opportunity for faster inferencing performance and real-time assistants similar to chatGPT while being cost-effective and running on a single GPU. Nonetheless, LLaMA was not adapted for instruction tasks with a Reinforcement Learning from Human Feedback (RLHF) training approach.

ChatLLaMA is an experimental chatbot interface that allows interaction with versions of Facebook's LLaMA. The 7 billion parameter model was fine-tuned with the Alpaca dataset.

Source: https://github.com/juncongmoo/chatllama