so it makes a checkpoint or a lora? Fine tune
so it makes a checkpoint or a lora? Fine tune
Fcheckpoint will be better
Fvery likely
GWhat do you mean by layers
�He means it's like the difference between Dreambooth vs making a Lora, if you were around earlier during the SD 1.5 days then that would make more sense
GI have been in this since 2021
GI see its like Dreambooth
Gor any of the major checkpoints/merges
Gits clear now
�Oh ok cool, yeah fine tuning Flux is just like dreambooth
Gi cant imagine how amazing flux will be in the coming weeks and months lmfao
GMJ all those are useless to us
GWhat strength do you use on the Lora? mine just finished on my 3090ti after 14hr 10 mins lol, and its over 2gb? why haha
DInteresting test!!!
�How's the quality of the Lora tho? Is it 
?
?Gits outrageous i cant believe what im seeing
Gits exactly what i programmed
Gwtf
Gfreaking sick af
�Awesome!
Gim so glad i have beasts at home, no overhead costs
GI have 3 GPUs
GCFG 3.5, Steps? and Lora Strength bro? what do you think is best in testing?
Fflux cfg 4 i use
F2gb true float save
F128 rank
G1 not 0.90 for strength? OK nice
Gthe lora works just trying to see the best settings, also you said to test other checkpoints?
Gepoch saves? i was wondering why I saw like 6 different ones in the lora folder
Gi only used the last one produced
Gi did the speed update on my 3090 now, do i need these settings on or off? and fp8 is not needed right? for 24gb VRAM card?
G
Ffp8 needed for 24gb
Fcompare checkpoints
Gi left it on
Gtesting to see the difference after i did the speed boost, i didnt do that last time and it took 14hrs
Gdid 32 images too
Gthis time doing 21 images
DWindows version is still running slow?
Fyep
Fi opened 2 threads on diffusers and pytorch
Fno replies yet
F
Foverfit but details - 1536 px - 256 lora rank - fine tuning will be a blast
F
F1024 vs 1536 - so much more details captured
B
BMay I ask why the descriptor breaks off in the middle and suddenly stops
Ihi guys 