for better output you need to train flux.1 dev with bfloat16 tensor bits
that is not possible with 16GB gpu
if you planed to rent cloud gpu then you can "https://github.com/ostris/ai-toolkit" fork , this opensource repo
that is providing high quality lora trainer
you can also get some hyper...
you can use mostly all opensource AI models from huggingface.co
here you can interact with millions of different AI models through inference and distributed gpu space
ML engineer are sharing their model with full access code base that help you to put out business by customizing them.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.