#Lucid_Vision

1 messages · Page 1 of 1 (latest)

atomic swan
drifting peak
# atomic swan help please

This extension is mildly annoying to setup.

  1. Firstly you need to download a vision model that you want to use.
    For example since you selected minicpm, you could try to put this in downloader field in model tab and download it:
    openbmb/MiniCPM-Llama3-V-2_5-int4

  2. Once it downloads, you need to go into:
    G:\AI\text-generation-webui-main2024\extensions\Lucid_Vision
    And open config.json

  3. Change fields:

"image_history_dir": "G:/AI/text-generation-webui-main2024/extensions/Lucid_Vision/ImageHistory/",

"minicpm_llama3_model_id": "G:/AI/text-generation-webui-main2024/models/openbmb_MiniCPM-Llama3-V-2_5-int4/",

  1. Restart the app and try again.
#

If you want to use other models, you need to download and manually setup their paths too.

atomic swan
#

we only work with LLAMA models with the suffix GGUF
28 GB RAM
because I have an old CPU model FX 8300
but I have a 1050 TI GTX and a 3050 RTX

is the option of good selected or change or should I forget about it?

drifting peak
#

I don't think this extension can work with GGUF models.
It seems to work only with models that can be loaded with Transformers library.

atomic swan
#

I don't know how to implement them, we managed to run only GGUF