BEST WEIGHT IS 0.6~0.8, DON'T HIGHER THAN 0.8, IT'S VERY VERY VERY TERRIBLE!!!
＝＝＝＝＝＝＝＝＝＝
WHERE I GENERATE IMAGES OR TRAIN LORAS
https://tensor.art/u/620420378505863684
TRAIN LORA & GENERATE IMAGES
https://pixai.art/@junkaiwang/artworks
TRAIN LORA & GENERATE IMAGES
https://civarchive.com/user/kaibear0428
PUBLISH LORA
https://civarchive.com/user/junkaiwang
GENERATE IMAGES
Description
FAQ
Comments (5)
Can you please leave metadata/prompt info on the generated images you post for future LORAs? Disappointing more and more people who make LORAs leave this info out these days.
Did you mean "Epochs" "Steps" "Clip Skip" ? I trained LORA use another website, HE didn't offer those data.
@kaibear0428 That's unfortunate, but I am talking mainly about samplers like DPM++ 2M Karras or Euler A, steps as in how many steps used to generate the image (20-30), what checkpoint was used to generate these images, and obviously the prompts.
But if it's out of your control and you don't exactly know what those are, then it can't be helped unless you find a service that can provide that info automatically.
when I upload pics, those data disappear, you can data you wants in other LORAs and PICs, the same sampler and steps.
@kaibear0428 That's fine, I was just curious as to what checkpoint and other settings were used to make those images since the style is unique enough to distinguish itself.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.