Kira Noir: KIRANOIR_1: kira_Lora128.safetensors

SD 1.5 Standard Model
Download

Info

Images: 114

Repeats: 20

Dimm: 128

Base

v1.5-pruned.ckpt

Client

Kohya_ss


LOG:
running training

  • num train images * repeats / Number of learning images x number of iterations: 2280

  • num reg images / Number of regularized images: 0

  • num batches per epoch / Batch size per 1 epoch: 1140

  • num epochs / epoch: 1

  • batch size per device / Batch size: 2

  • total batch size (including parallel learning and gradient sum: 2

  • gradient accumulation steps / Number of steps to sum gradients = 1

  • total optimization steps / Number of learning steps: 1140

Generated examples are with Realistic Vision V1.3

First ever trained LoRA.

***NSFW by nature, be deliberate with sfw prompts and hope for it to generate with clothes ***

check Model Version description for training log