LoRA Finetuning - Aff4n20/wuerstchen-ancient-coins

This pipeline was finetuned from warp-ai/wuerstchen-prior on the Aff4n20/ancient-coin-dataset dataset. Below are some example images generated with the finetuned pipeline using the following prompts: ['inscription, IMP AVG DIVI F; bare head of Augustus left; in front palm; behind, winged caduceus']:

Pipeline usage

You can use the pipeline like so:

from diffusers import DiffusionPipeline
import torch

pipeline = AutoPipelineForText2Image.from_pretrained(
                "warp-ai/wuerstchen", torch_dtype=torch.float16
            )
# load lora weights from folder:
pipeline.prior_pipe.load_lora_weights("Aff4n20/wuerstchen-ancient-coins", torch_dtype=torch.float16)

image = pipeline(prompt=prompt).images[0]
image.save("my_image.png")

Training info

These are the key hyperparameters used during training:

• LoRA rank: 4
• Epochs: 19
• Learning rate: 0.0001
• Batch size: 1
• Gradient accumulation steps: 4
• Image resolution: 512
• Mixed-precision: fp16

More information on all the CLI arguments and the environment are available on your wandb run page.