Closed mialana closed 1 month ago
Thanks for your interest. The code of LLM inference is at https://github.com/OpenRobotLab/GRUtopia/blob/6e20ccdff67faa4a485c93528b925829477a6ffb/grutopia/npc/llm_caller.py#L82
You can modify the api_base_url
if you are using an local llm server or replace it with inference code of your local LLM.
Feel free to reach out if you have other questions.
Ok, thank you. And if I want to host the WebUI without using Docker, is there a quick way to do this? I have Isaac SIM locally.
And if I want to host the WebUI without using Docker, is there a quick way to do this?
@mialana Currently there's no quick way to realize that, since nginx is required to serve the WebUI frontend, and we believe that setting it up with docker is a simpler way for most of the researchers. You can refer to Dockerfile for the the settings if a local nginx is more suitable for you. And you can start the WebUI backend by
python ./webui/main.py
By the way, the WebUI process in container is able to communicate to locally installed Isaac Sim, but some modification is necessary to make WebRTC work properly:
You can do it against the locally installed Isaac Sim, and then start WebUI process only in the docker container. They should be able to work with each other.
您好。I am curious if there are plans to eventually make NPC interaction available in the local GRUtopia environment? Thanks!