Taggui feedback and support

[WarAnakin]

Thank you for creating this amazing tool, it has helped me and my team bring quite a few different projects to a successful finish. These projects vary in size and scope with some of the datasets reaching well above 2k images and captions.
We train premium stable diffusion models (checkpoints and loras) for a number of companies and individuals. If you are an active user of stable diffusion, then most likely you have used our models to generate images.
Let me know if you have a patreon or a paypal support account.

[jhc13]

Thank you so much for your kind words! I have always developed the program with large datasets in mind, and I am glad that it has been useful for you and your team.

As for your generous offer of support, while I do not have a Patreon or a Paypal page set up, I do have a Ko-fi page that can accept donations.

Thank you again for the feedback and support! I really appreciate it, and I will continue to work on improving TagGUI. Feel free to let me know if you have any additional feedback or suggestions.

[WarAnakin]

I do have suggestions and so far I am glad to see that somehow you caught wind of the selection inversion process, it was few months ago that i openly suggested this idea in the community.
Seeing that the datasets i usually have to deal with are quite large and varied, i have a few ideas on how captioning can become a more effective process. We can't use auto-captioning because the system cant give us the accuracy and complexity that is requested of us in some of these projects.

[jhc13]

I agree that the current auto-captioning models are not quite good enough yet to replace manual captioning. They are improving fast though, and CogVLM is already very good. I would love to hear your ideas on improving the captioning process, if you would like to share them.

The best way to suggest an idea would be through creating a discussion or issue on GitHub (although the selection inversion feature was suggested to me on Discord).

[WarAnakin]

I know. I'm the one that told him and then he told you. Come find me on Discord, my account has the same name as here.

[BenDes21]

Hi, thanks for your tool!! Do you know if they're is a way to ron CogVLM with low ram GPU ( rtx 4080 12GB ) Thanks

[jhc13]

Do you know if they're is a way to ron CogVLM with low ram GPU ( rtx 4080 12GB ) Thanks

You'll need a slightly bigger GPU for CogVLM in 4-bit (around 14 GB will be enough).