onsite trainer QoL

trainer needs ways to test loras more (test it with not puplishing it to all so the maker can see if its works as he want! not any one have a 4090) and use more of the epochs and even a way to pick more as 1 epoch and lets train with this the same lora again as options!

i try the onsite trainer for a lora to made more of some pics i liked but this trainer is to bad at the steps and anything!

the first step is fine who you pick pics you cant the ai to train on/with!

i missing a way to force the lora to use a Tag/key as trigger so you can train the AI to only use things with yout trigger word. (i dont finde a way to make this sry)

then moste of the explanation at least for me are not realy good to understand waht dose waht!

after the “training” is done you can choise only 1 epoch! you cant say take this and this and train again!

and i cant finde the option who i chose the weight the lora will use ( i mean this -10.0 to 20.0 or other numbers. most loras have only -2.0 to 2.0. i must missed this).

after this you can say waht yout lora site will say and show.. not more.

after this you can ONLY publishe! not try to test your OWN lora to see if its work!

and i finde a option to train this lora on site AGAIN! to make V2 better or even train it to make V1 better and work like you want it!

I am btw open for tips&help for make a lora on site with the on site trainer!

if some one have done good one with the onsite trainer etc who want help me .. chat or pn me. if this works on the site.