Software Engineering Courses (SECourses)•16mo ago

do you use system prompt? if so what is it

FFurkan Gözükara SECourses do you use system prompt? if so what is it

Maxivi•9/8/24, 4:10 PM

just caption the image with details, colors, makeup, items, objects, emotions, art style, drawing style and objects but do not add any description or comment. do not miss any item in the given image, very simple sentance.

Maxivi•9/8/24, 4:10 PM

they have a demo interface

Maxivi•9/8/24, 4:10 PM

https://internvl.readthedocs.io/en/latest/get_started/local_chat_demo.html

Maxivi•9/8/24, 4:12 PM

@Dr. Furkan Gözükara its quite nice actually, able to capture details better as well. i've tried longer prompts too works quite nice

Maxivi•9/8/24, 4:12 PM

i think it would be better than joy and cog for style training too

FFurkan Gözükara SECourses you need to generate if folder missing

xlads•9/8/24, 4:13 PM

just create a folder titled Umet in the model files inside of swarmui?

Dreamer@TDFilmStudio•9/8/24, 4:13 PM

@Dr. Furkan Gözükara Yes please see if InternVL2 is possible

Maxivi•9/8/24, 4:13 PM

it crazy that the 8b models outperforms many of the others

Maxivi•9/8/24, 4:13 PM

26b 4bit it possilbe as well, its the best one from my testing

MMaxivi just caption the image with details, colors, makeup, items, objects, emotions, a...

Furkan Gözükara SECoursesOP•9/8/24, 4:14 PM

they lack Gradio Demo

Xxlads just create a folder titled Umet in the model files inside of swarmui?

Furkan Gözükara SECoursesOP•9/8/24, 4:14 PM

unet

Furkan Gözükara SECoursesOP•9/8/24, 4:14 PM

exactly as that

Maxivi•9/8/24, 4:14 PM

they have streamlit demo

Furkan Gözükara SECoursesOP•9/8/24, 4:15 PM

their demo down

Furkan Gözükara SECoursesOP•9/8/24, 4:15 PM

Calling st.rerun() within a callback is a no-op.

Maxivi•9/8/24, 4:16 PM

https://internvl.opengvlab.com/

Maxivi•9/8/24, 4:16 PM

this one? its working for me

Furkan Gözükara SECoursesOP•9/8/24, 4:16 PM

testing here

Furkan Gözükara SECoursesOP•9/8/24, 4:16 PM

https://huggingface.co/spaces/OpenGVLab/InternVL

InternVL - a Hugging Face Space by OpenGVLab

Maxivi•9/8/24, 4:16 PM

both work

Maxivi•9/8/24, 4:16 PM

even for me

Furkan Gözükara SECoursesOP•9/8/24, 4:17 PM

Maxivi•9/8/24, 4:18 PM

Seems pretty good!

Furkan Gözükara SECoursesOP•9/8/24, 4:18 PM

Furkan Gözükara SECoursesOP•9/8/24, 4:18 PM

now i will compare generated images on swarmui

FFurkan Gözükara SECourses Click to see attachment

Maxivi•9/8/24, 4:19 PM

i think in terms of training, when you train this many tokens for a person its not good, but the simplified but accurate ones from internvl gave me quite good results

MMaxivi i think in terms of training, when you train this many tokens for a person its n...

Furkan Gözükara SECoursesOP•9/8/24, 4:19 PM

well i think for training a person no captions better :d

FFurkan Gözükara SECourses well i think for training a person no captions better :d

Maxivi•9/8/24, 4:20 PM

depends, yours are usually overtrained

xlads•9/8/24, 4:26 PM

okay i made the unet file inside of models then restarted/updated but still getting same error saying "All available backends failed to load the model"