lora-training

This is a LoRA derived from black-forest-labs/FLUX.1-dev.

The main validation prompt used during training was:

A fashionable woman wearing oversized white sunglasses and orange lipstick. woman, fashion, sunglasses, oversized glasses, white sunglasses, orange lipstick, makeup, portrait, close-up, monochrome, dark hair, straight hair, chic, stylish, modern, glamour, glossy, high contrast, minimalistic background, beauty, contemporary, bold, accessory, trendy, cosmetic, facial expression, luxurious, elegant, sophisticated, high fashion.

Validation settings

• CFG: 3.5
• CFG Rescale: 0.0
• Steps: 20
• Sampler: None
• Seed: 42
• Resolution: 1024x1024

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

You can find some example images in the following gallery:

Prompt

unconditional (blank prompt)

Negative Prompt

'

Prompt

your prompt to validate on

Negative Prompt

'

Prompt

another prompt to validate on

Negative Prompt

'

Prompt

A fashionable woman wearing oversized white sunglasses and orange lipstick. woman, fashion, sunglasses, oversized glasses, white sunglasses, orange lipstick, makeup, portrait, close-up, monochrome, dark hair, straight hair, chic, stylish, modern, glamour, glossy, high contrast, minimalistic background, beauty, contemporary, bold, accessory, trendy, cosmetic, facial expression, luxurious, elegant, sophisticated, high fashion.

Negative Prompt

'

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

Training settings

• Training epochs: 142
• Training steps: 2000
• Learning rate: 8e-07
• Effective batch size: 20
  • Micro-batch size: 10
  • Gradient accumulation steps: 2
  • Number of GPUs: 1
• Prediction type: flow-matching
• Rescaled betas zero SNR: False
• Optimizer: AdamW, stochastic bf16
• Precision: Pure BF16
• Xformers: Not used
• LoRA Rank: 16
• LoRA Alpha: None
• LoRA Dropout: 0.1
• LoRA initialisation style: loftq

Datasets

poc1_lundlund_hugocomte

• Repeats: 0
• Total number of images: 280
• Total number of aspect buckets: 1
• Resolution: 512 px
• Cropped: True
• Crop style: center
• Crop aspect: square

Inference

import torch
from diffusers import DiffusionPipeline

model_id = 'black-forest-labs/FLUX.1-dev'
adapter_id = 'hmpark/lora-training'
pipeline = DiffusionPipeline.from_pretrained(model_id)
pipeline.load_lora_weights(adapter_id)

prompt = "A fashionable woman wearing oversized white sunglasses and orange lipstick. woman, fashion, sunglasses, oversized glasses, white sunglasses, orange lipstick, makeup, portrait, close-up, monochrome, dark hair, straight hair, chic, stylish, modern, glamour, glossy, high contrast, minimalistic background, beauty, contemporary, bold, accessory, trendy, cosmetic, facial expression, luxurious, elegant, sophisticated, high fashion."


pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu')
image = pipeline(
    prompt=prompt,
    num_inference_steps=20,
    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=3.5,
).images[0]
image.save("output.png", format="PNG")