Tipo
LORA
Data di pubblicazione
2024-10-12
Modello Base
Flux.1 D
Parola Chiave
Copia
Vedi di più

Heres two models. On i start create with 64/32 dim/alpha, but learning is but unfortunately, the training was interrupted, so I managed to download only 14 epochs out of 20.
The second version has already been successfully trained (20\20). But in it I tried to use the 4/32 dim/alpha ratio. This is how this ratio was recommended in the civitai training system itself.