Curious about the multi-turn conversation demo

[chris5zk]

Hi, thank you for making this great work!

I run the demo code chat.py using the following script: CUDA_VISIBLE_DEVICES=0 python chat.py --version='xinlai/LISA-13B-llama2-v1' --precision='fp16' --load_in_4bit on my GPU. It present the USER, ASSISTANT conversation as you demonstrate in your README.

I am curious about if the model keep the conversation history or not? I try to make a conversation with the model but I am not sure the response from it is depend on the previous dialogue or just the current input prompt?

Thanks!