Comment by numlocked

Did you run the fine-tuning on LLaMA yourselves based on the 52k examples from Alpaca? Or is there a 7B pre-trained alpaca model out there that you grabbed?