jhc13 / taggui

Tag manager and captioner for image datasets
GNU General Public License v3.0
707 stars 33 forks source link

Previously downloaded models #268

Closed Yp-pro closed 1 month ago

Yp-pro commented 1 month ago

Hello,

I'm trying to follow this:

If you have a local directory containing previously downloaded models, you can set it in File -> Settings to include the models in the model list.

https://github.com/jhc13/taggui?tab=readme-ov-file

I downloaded a model, put the files into my directory, set the path up in program, restarted the program.

But the program starts to download the model after clicking "Start Auto Captioning" anyway.

What am I doing wrong? Please help.

P.S.: BTW I don't see my model in the dropdown menu, so I can't select it

jhc13 commented 1 month ago

How did you structure the files? The added model should be visible separately in the dropdown list (at the top).

Yp-pro commented 1 month ago

How did you structure the files? The added model should be visible separately in the dropdown list (at the top).

image

And set the path to Models in Settings

jhc13 commented 1 month ago

Is that all the files? There should be more files, like the config file.

Yp-pro commented 1 month ago

What the config files? I downloaded the files from here: https://huggingface.co/Salesforce/blip2-opt-2.7b/tree/main Should I take something else except .safetensors?

jhc13 commented 1 month ago

You will need the config.json file at least. I'm not sure about the others.

Yp-pro commented 1 month ago

You will need the config.json file at least. I'm not sure about the others.

It works. Thank you!