Question regarding this model
Thank you so much for developing this model it seems so good and I'd like to know if i want to run it locally as the demo space (https://huggingface.co/spaces/ningshanwutuobang/panda_gpt_llama_cpp ) , is it enough to download this ggml file or should i download the below file which includes the weights too? openllmplayground/pandagpt_13b_max_len_400
There are alreadly codes in the file https://huggingface.co/spaces/ningshanwutuobang/panda_gpt_llama_cpp/blob/17e82fc693857d629d7064e5595dcde93b5011ef/app.py#L7 to download it.
So i can't use the GGML file on its own in koboldcpp?
The image input required ImageBind model and a projection layer from of PandaGPT. If you use text only, you can use the ggml file directly. I am not sure for the image input for koboldcpp. I think it is different.
I only want to use text generation features. Will i get the same output text results as the demo, without using openllmplayground/pandagpt_13b_max_len_400?
It could be the same, but the results will depend on the temperature and random seed.
Thank you so much for this model and your support. I'm downloading this model and hope to be able to use it.