flux-training-seuss-lora

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

The main validation prompt used during training was:

A photograph of Dr. Seuss riding in a horse-drawn carriage

Validation settings

• CFG: 3.5
• CFG Rescale: 0.0
• Steps: 15
• Sampler: None
• Seed: 42
• Resolution: 1024

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

blurry, cropped, ugly

Prompt

Dr. Seuss during a book signing event, seated at a table with an open book and pen in hand, his characteristic white beard, clear-rimmed glasses, and whimsical bow tie complementing his calm, attentive expression, all within the literary setting of a bookstore, reflecting his enduring connection with readers and the joy his work brought to many.

Negative Prompt

blurry, cropped, ugly

Prompt

Anime picture of famed author Dr. Seuss in a Studio Ghibli style

Negative Prompt

blurry, cropped, ugly

Prompt

Dr. Seuss in a leather jacket riding a Harley Davidson Motorcycle

Negative Prompt

blurry, cropped, ugly

Prompt

Famous author Dr. Seuss holding a chainsaw while riding around on a unicycle, vintage TV still from the Dick Van Dyke show

Negative Prompt

blurry, cropped, ugly

Prompt

A photograph of Dr. Seuss riding in a horse-drawn carriage

Negative Prompt

blurry, cropped, ugly

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

Training settings

• Training epochs: 0
• Training steps: 100
• Learning rate: 0.0008
• Effective batch size: 16
  • Micro-batch size: 4
  • Gradient accumulation steps: 4
  • 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

default_dataset_arb

• Repeats: 100
• Total number of images: 4
• Total number of aspect buckets: 3
• Resolution: 1.5 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None

default_dataset

• Repeats: 100
• Total number of images: 3
• Total number of aspect buckets: 1
• Resolution: 1.048576 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square

default_dataset_512

• Repeats: 100
• Total number of images: 4
• Total number of aspect buckets: 1
• Resolution: 0.262144 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square

default_dataset_576

• Repeats: 100
• Total number of images: 4
• Total number of aspect buckets: 1
• Resolution: 0.331776 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square

default_dataset_640

• Repeats: 100
• Total number of images: 4
• Total number of aspect buckets: 1
• Resolution: 0.4096 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square

default_dataset_704

• Repeats: 100
• Total number of images: 4
• Total number of aspect buckets: 1
• Resolution: 0.495616 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square

default_dataset_768

• Repeats: 100
• Total number of images: 3
• Total number of aspect buckets: 1
• Resolution: 0.589824 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square

default_dataset_832

• Repeats: 100
• Total number of images: 3
• Total number of aspect buckets: 1
• Resolution: 0.692224 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square

default_dataset_896

• Repeats: 100
• Total number of images: 3
• Total number of aspect buckets: 1
• Resolution: 0.802816 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square

default_dataset_960

• Repeats: 100
• Total number of images: 3
• Total number of aspect buckets: 1
• Resolution: 0.9216 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square

Inference

import torch
from diffusers import DiffusionPipeline

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

prompt = "A photograph of Dr. Seuss riding in a horse-drawn carriage"

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=15,
    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")