>>11325534
>Idk if I'm going to go back to training loras yet
Fair enough
>datasets
lasto and soranamae look easy to train in this manner where you reinforce the tag, nezunezu a bit iffy maybe, might be easier to train their e621 tag (nezumi (artist)) and for the other two it looks like you wouldn't get any help from inherent knowledge in the model, unfortunately
>kunaboto lora
I did, yeah, thanks!
>>11325632
very nice