This is a test of training a MOSS LoRA on-site. I am uploading the result in order to test it on-site, I only manage to generate an ugly pale/yellowish result locally :-(

All the showcase pictures has been generated on-site, this LoRA is only a test, since i had a big dataset ready, i just tested it for MOSS ^^;

It was a tagged dataset, so, it should be able to handle tag prompt instead of NLP. No trigger word, maybe except "masterpiece, best quality, highres"

Trained on Base, works nicely with a Turbo LoRA (check showcase picture for usage).