2nd attempt at lora training, again any feedback is very much appreciated.

V2 improved on generated eyes though not perfect, may still get some quirky results, recommeded strength at 0.6.

Will be working on V3, seems like overfitted.

V3 update:

Please avoid the previous 2 versions. V3 is much more better. Still having some problems with the eyes. Will try to fix it in V4. V3 does not have trigger word.