What AI says about Stanford Alpaca

Alpaca is a small instruction-following language model derived from Meta's LLaMA 7B, fine-tuned on 52k instruction-following demonstrations to perform comparably to OpenAI's text-davinci-003 on single-turn tasks while being inexpensive and easy to reproduce (under $600). The project provides its training recipe and data, with plans to release the model weights, and offered an interactive demo (now disabled) to help researchers study the model's behavior and limitations. Usage is restricted to academic research; commercial use is prohibited due to licensing and safety considerations.