search

jhc13 / taggui

Tag manager and captioner for image datasets
GNU General Public License v3.0
774 stars 37 forks source link

[Model Request]: Phi-3-vision-128k-instruct #246

Closed BenDes21 closed 4 months ago

BenDes21 commented 4 months ago

Hi there, is it possible to add this interesting model for the captionning ?

Infos : https://huggingface.co/microsoft/Phi-3-vision-128k-instruct

( I guess finetune like https://huggingface.co/Desm0nt/Phi-3-HornyVision-128k-instruct will be able to be use by setting a model folder )

Thanks

jhc13 commented 4 months ago

This was already requested in #158.

( I guess finetune like https://huggingface.co/Desm0nt/Phi-3-HornyVision-128k-instruct will be able to be use by setting a model folder )

Instead of downloading manually, you could just put Desm0nt/Phi-3-HornyVision-128k-instruct in Model and it would work (if support for the base model were to be added).

jhc13 commented 4 months ago

This model has been added in v1.30.0.

Seedmanc commented 3 months ago

Why does it require an RTX 30xx to run? Since when models have hardware requirements other than the VRAM size?