Weight: 0.1-0.9
Vague and impractical content, different Checkpoint can get different results, recommended to try.
Description
FAQ
Comments (5)
The metadata for this model reports 1250 images were used but there are only a total of 154 instances of keyword tags present. What is your strategy for when to apply tags to an image and when not to?
This is a very old model, and as far as I remember, it was trained with only 25-35 images. I did not keep the original data because I felt that the model did not meet my expectations.
@MetroCat the safetensors_util tool from here is nice: https://github.com/by321/safetensors_util Or just use vim and deal with the ugly jumbled mess. In comfy the rgthree custom nodes includes a couple of LoRA loaders. If you right click the loaded LoRA you can view a list of tags, highlight, and copy/paste them.
The particular part of the metadata reads:
"ss_dataset_dirs": {"50_Dryad": {"n_repeats": 50, "img_count": 25}}
which means 25 unique images were used and repeated 50 times. 1250 is 25×50.