Trained on 65 images with detailed captions (...took...so..much...time..captioning)
Trained at 704 resolution, clip skip 2, no vae. 65 images at 10 repeats for 5 epochs at batch 5- which led to 650 steps. Unet 0.00035, TE 0.00012
Trained on 65 images with detailed captions (...took...so..much...time..captioning)
Trained at 704 resolution, clip skip 2, no vae. 65 images at 10 repeats for 5 epochs at batch 5- which led to 650 steps. Unet 0.00035, TE 0.00012