This is a Chinese instruction-tuning lora checkpoint based on llama-7B(2epoch) from this repo's work
-