what is your batch size
what is your batch size
N
Fsend me your json file as prrivatre message
Fyou are not using lora tab
Fnow extract config from zip file again
Ffresh
Fload into lora tab fresh
Fthen set your files
Fit will work
Fi looked your config file
Noh damn
Nkohya is new for me, my bad
F
Fi am making video hopefully today
Nyou can add a first advice " Change Tabs to LORA!")))
F i am gonna do nowFbut recording video and will show all
Nyey! but speed is lower than expected. I guess 4060ti is not the best one for training
Fare you using shared vram
Fspeed will get better
Fwait like 150 steps to see final speed
NI guess not
Nbut my ram is full I guess with model and clips
Fye not using
Fwait 150 steps
Fand show me speed
Ndidn't change much
N@Furkan Gözükara SECourses is there any method to train lora by parts, because not always there is possibility to have pc turned on?
N200 steps same
Fok
Fso to speed up you have 2 options
Fturn off t5 text encoder masking attention
Fvery little quality loss
Fspeed up signficiant
Fif still slow reduce training resolut
Flike 768x768
Fthis will hve more impact on quality
Ngot it. Will be trying new parameters tomorrow. Will lyk results
Nthanks
Fyou are welcome
XI was wondering, do warmup steps damage a lora?
Fwarmup steps means lower or higher LR
Fmay damage
YI am making a LoRA of a realistic character. The training data contain face swapped images and most have too much glare around the eyes. If I were to include a caption about eyes having too much glare, (say "eye glare") would the LoRA train to only keep the eye glare if "eye glare" were in the prompt, and otherwise not include the eye glare if it were not mentioned?
Ftraining for flux?
Yyes
Fflux already has automatic inner caption system
Fbut you can add eye glare to caption
Fafter training dont use in prompt
Fmay reduce eye glare effect
Fafter training