sd3-lora-training-v2

This is a standard PEFT LoRA derived from stabilityai/stable-diffusion-3-medium-diffusers.

The main validation prompt used during training was:

ethnographic photography of teddy bear at a picnic

Validation settings

• CFG: 5.0
• CFG Rescale: 0.0
• Steps: 30
• Sampler: None
• Seed: 42
• Resolution: 1024x1024

Note: The validation settings are not necessarily the same as the training settings.

The text encoder was not trained. You may reuse the base model text encoder for inference.

Training settings

• Training epochs: 89
• Training steps: 2960
• Learning rate: 8e-07
• Effective batch size: 2
  • Micro-batch size: 1
  • Gradient accumulation steps: 2
  • Number of GPUs: 1
• Prediction type: flow-matching
• Rescaled betas zero SNR: False
• Optimizer: adamw_bf16
• Precision: bf16
• Quantised: No
• Xformers: Not used
• LoRA Rank: 16
• LoRA Alpha: None
• LoRA Dropout: 0.1
• LoRA initialisation style: default

Datasets

Pal_BLIP

• Repeats: 0
• Total number of images: 73
• Total number of aspect buckets: 1
• Resolution: 1.048576 megapixels
• Cropped: True
• Crop style: center
• Crop aspect: square

Inference

import torch
from diffusers import DiffusionPipeline

model_id = 'stabilityai/stable-diffusion-3-medium-diffusers'
adapter_id = 'zwloong/sd3-lora-training-v2'
pipeline = DiffusionPipeline.from_pretrained(model_id)
pipeline.load_lora_weights(adapter_id)

prompt = "ethnographic photography of teddy bear at a picnic"
negative_prompt = 'blurry, cropped, ugly'
pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu')
image = pipeline(
    prompt=prompt,
    negative_prompt=negative_prompt,
    num_inference_steps=30,
    generator=torch.Generator(device='cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu').manual_seed(1641421826),
    width=1024,
    height=1024,
    guidance_scale=5.0,
).images[0]
image.save("output.png", format="PNG")