Little Nemo as he appeared in the 1989 anime movie Little Nemo: Adventures in Slumberland.
His default clothes are generated with this tag: white sleeping jumpsuit
Enjoy!
Description
More epochs, bigger batch size for training, improved and more accurate LORA
Comments (5)
Maybe it's just me, but I think "nemo \(littlenemo)\" causes problems in Forge Neo and therefore the lora doesn't work.
That's strange. I stopped using Forge Neo a number of months ago in favor of ComfyUI. I just tested my LORA a little more without the "nemo \(littlenemo)\" tag in the positive prompt and I can still get a good image of Nemo made. Try ignoring the tag that we mentioned in quotation marks and see if generation can work.
@ArtArtArt11 I was wondering about ComfyUI for some time (about three moths, maybe) - did they nerf the CLIP support on LoRAs? Cause my older ones I've trained (ILL and Pony, not posted, except one), with trained text encoder and normalization layer don't seem to give the results I had before.
@mikmod nerf CLIP support of LORAs? I am not aware of that sort of thing. Nowadays I just stopped using ILL models as I've fully moved on to Anima.
@ArtArtArt11 I've noticed the character I've trained wasn't 'represented' as well as before with the same workflow I've used then. That's why I thing the ComfyUI devs could have dropped the support for such format I've used (within the Lora-Easy-Training application). This training 'script' doesn't support Anima though.