Download
1 variant available
1670 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
4K0 1 2 3 4 5 6 7 8 9K
1.3K0 1 2 3 4 5 6 7 8 9.0 1 2 3 4 5 6 7 8 9K
(36)
Apr 2, 2025
trained based on noobai-xl eps v1.1
2540 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
1.5K0 1 2 3 4 5 6 7 8 9.0 1 2 3 4 5 6 7 8 9K
313K0 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9K
License:
NoobAI LicenseMithra (dragon village/dragon village m)
___________________________________________________
V1.0 (noobxl_eps)
triggers: dvmithra, dragon, two tone body
optional: bracelet, loincloth, headgear, chest accessory, tail accessory, wings, tail, scales
___________________________________________________
V1.0 (pony_xl)
triggers: Mithra, dragon
optional: bracelet, loincloth, wings, tail
Tags that are trained but have little control: headgear, chest accessory
known issues: Sometimes a "human" is generated because of the body color of this character. Try using "anthro" to the positive or "human" to the negative.
___________________________________________________
This is the JSON file containing the configuration used for LoRA training.
https://drive.proton.me/urls/WAJ2FF3PG0#8xf2yPDr1iK3
Additionally, metadata is stored within the .safetensors file, so you can inspect its contents by opening it with a program like Notepad.
There's no special secret to these settings; they are largely similar to what others using Prodigy would employ. Ultimately, the dataset is the primary determinant of training outcomes. I'd like to attempt training other DVM characters as well, but it's challenging due to the limited number of available images for them. Mithra was an exception, as his significant popularity made it feasible.