# Training summary

This model was trained by using 40 manually sampled images for 10 epochs, with 120 iterations per epoch.

The following is an excerpt of some of the learning commands by sd-script for reference

train_network.py --pretrained_model_name_or_path="JosephusCheung/ACertainty" --prior_loss_weight=1.0 --enable_bucket --min_bucket_reso=384 --max_bucket_reso=1280 --train_batch_size=12 --learning_rate=1e-4 --text_encoder_lr 5e-5 --use_8bit_adam --xformers --mixed_precision=fp16 --save_every_n_epochs=10 --save_model_as=safetensors --clip_skip=2 --seed=42 --color_aug --flip_aug --network_module=networks.lora --resolution=768,512 --network_dim 256 --max_train_epochs 4 --shuffle_caption

# Problems
This model has a major flaw. Ryo's trademark two hair clips collapse frequently. Also, the directions are scattered like musical ideas. How can this be fixed? looking for ideas.