[Return]

Report a post

Preview
I'm currently working on expanding the dataset for the Lora. I've set the arbitrary goal of 42 images per bladed weapon type that (in an ideal world) I want the Lora(s) to be capable of reproducing. 42 because its a nice number and also because the dataset that was used to create the pussystab_v1 lora contained exactly 42 pics of knives and imo knives so far work the best. So out of 8 planned weapon types I've got at least this category already covered...
Post number No.2732
Board Artificial Intelligence
Optional. Describe what's wrong with it.