64枚の画像で学習済み

{
  "engine": "kohya",
  "unetLR": 0.0003,
  "clipSkip": 1,
  "loraType": "lora",
  "keepTokens": 3,
  "networkDim": 32,
  "numRepeats": 4,
  "resolution": 1024,
  "lrScheduler": "cosine",
  "minSnrGamma": 0,
  "noiseOffset": 0,
  "targetSteps": 1280,
  "enableBucket": true,
  "networkAlpha": 32,
  "optimizerType": "Adafactor",
  "textEncoderLR": 0,
  "maxTrainEpochs": 20,
  "shuffleCaption": true,
  "trainBatchSize": 4,
  "flipAugmentation": true,
  "lrSchedulerNumCycles": 1
}