LoRA training using thousands of photos I've been collecting on Instagram one by one for the last couple of years. Specializing in sexy, handsome, white muscle men.
My personal Instagram : Lyahn.here
**Highly(or only) recommended to use with Realistic Stock Photo checkpoints for the best results.
Here are some example Instagram images I used for training:
Recommend prompt : A portrait photograph of an extremely handsome caucasian white man model, male focus, 1boy, solo, beard, model, attractive young man, sexy, hot, seductive, gay, realistic, details, photo, short hair
Recommend negative prompt : (Women, girl), deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime, text, cropped, out of frame, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck
Description
This version utilized a higher quality dataset to achieve better results.
Please note that it should only be used with Realistic Stock Photo .
Training Params:
{ "unetLR": 0.0005, "clipSkip": 1, "loraType": "lora", "keepTokens": 0, "networkDim": 200, "numRepeats": 1, "resolution": 2048, "lrScheduler": "cosine_with_restarts", "minSnrGamma": 5, "noiseOffset": 0.1, "targetSteps": 4123, "enableBucket": true, "networkAlpha": 200, "optimizerType": "Adafactor", "textEncoderLR": 0.00005, "maxTrainEpochs": 19, "shuffleCaption": true, "trainBatchSize": 4, "flipAugmentation": true, "lrSchedulerNumCycles": 3 }