FranzWilhelmSeiwert-Flux-LoKr-4e-4

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

No validation prompt was used during training.

None

Validation settings

• CFG: 3.0
• CFG Rescale: 0.0
• Steps: 20
• Sampler: FlowMatchEulerDiscreteScheduler
• Seed: 42
• Resolution: 1024x1024
• Skip-layer guidance:

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

In the style of a Raphael oil painting, Three figures in red and white religious attire, with one seated at a table holding a magnifying glass, and two standing figures behind. The table is covered with a red cloth and holds an open book and a silver bell. Dark background and muted color palette.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Raphael oil painting, A knight on a white horse is spearing a dragon lying on the ground. The knight wears armor and a blue cape, while a woman in a red dress stands in the background beside a rock formation. The setting includes tall trees and a distant cityscape.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Raphael oil painting, A bearded man wearing a black robe and cap sits at a table holding papers in one hand. An apple rests on the table alongside a book with a ring visible on his finger. The background is plain and neutral.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Raphael oil painting, A seated figure in a blue robe and red dress holds a book, surrounded by two young, unclothed children in a natural setting with trees and mountains in the background. One child holds a bird while the other reaches out towards it. The setting includes a rock and a grassy landscape.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Raphael oil painting, A scholar-alchemist in flowing robes stands amid glass vessels and astronomical instruments, while light streams through a Gothic window. A mechanical armillary sphere sits prominently on a wooden table, while an assistant in the background tends to a burning crucible.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Raphael oil painting, Neptune rises from turbulent waters on the steps of Venice's St. Mark's Basilica, offering a golden ring to a figure representing the Maritime Republic. Merchants in Renaissance dress observe from gondolas, while angels hold scrolls of maritime law above.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Raphael oil painting, Aristotle and Plato walk through a Renaissance medicinal garden, discussing a dissected flower. Young apprentices sketch botanical specimens nearby, while in the background, monks tend to rows of healing herbs beneath a pergola covered in grape vines.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Raphael oil painting, Angels and scholars share a vast library with soaring Renaissance architecture, where celestial maps float in mid-air. Some angels point to globes showing undiscovered continents, while others transcribe from books bound in supernatural light. A telescope made of gold and ivory points through an open dome to the stars.

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: 18
• Training steps: 4500
• Learning rate: 0.0001
  • Learning rate schedule: constant
  • Warmup steps: 200
• Max grad norm: 2.0
• Effective batch size: 3
  • Micro-batch size: 3
  • Gradient accumulation steps: 1
  • Number of GPUs: 1
• Gradient checkpointing: True
• Prediction type: flow-matching (extra parameters=['shift=3', 'flux_guidance_mode=constant', 'flux_guidance_value=1.0', 'flow_matching_loss=compatible'])
• Optimizer: adamw_bf16
• Trainable parameter precision: Pure BF16
• Caption dropout probability: 10.0%

LyCORIS Config:

{
    "algo": "lokr",
    "multiplier": 1.0,
    "linear_dim": 10000,
    "linear_alpha": 1,
    "factor": 16,
    "apply_preset": {
        "target_module": [
            "Attention",
            "FeedForward"
        ],
        "module_algo_map": {
            "Attention": {
                "factor": 16
            },
            "FeedForward": {
                "factor": 8
            }
        }
    }
}

Datasets

fws-512

• Repeats: 10
• Total number of images: 10
• Total number of aspect buckets: 5
• Resolution: 0.262144 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

fws-1024

• Repeats: 10
• Total number of images: 10
• Total number of aspect buckets: 8
• Resolution: 1.048576 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

fws-512-crop

• Repeats: 10
• Total number of images: 10
• Total number of aspect buckets: 1
• Resolution: 0.262144 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square
• Used for regularisation data: No

fws-1024-crop

• Repeats: 10
• Total number of images: 10
• Total number of aspect buckets: 1
• Resolution: 1.048576 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square
• Used for regularisation data: No

Inference

import torch
from diffusers import DiffusionPipeline
from lycoris import create_lycoris_from_weights


def download_adapter(repo_id: str):
    import os
    from huggingface_hub import hf_hub_download
    adapter_filename = "pytorch_lora_weights.safetensors"
    cache_dir = os.environ.get('HF_PATH', os.path.expanduser('~/.cache/huggingface/hub/models'))
    cleaned_adapter_path = repo_id.replace("/", "_").replace("\\", "_").replace(":", "_")
    path_to_adapter = os.path.join(cache_dir, cleaned_adapter_path)
    path_to_adapter_file = os.path.join(path_to_adapter, adapter_filename)
    os.makedirs(path_to_adapter, exist_ok=True)
    hf_hub_download(
        repo_id=repo_id, filename=adapter_filename, local_dir=path_to_adapter
    )

    return path_to_adapter_file
    
model_id = 'black-forest-labs/FLUX.1-dev'
adapter_repo_id = 'davidrd123/FranzWilhelmSeiwert-Flux-LoKr-4e-4'
adapter_filename = 'pytorch_lora_weights.safetensors'
adapter_file_path = download_adapter(repo_id=adapter_repo_id)
pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
lora_scale = 1.0
wrapper, _ = create_lycoris_from_weights(lora_scale, adapter_file_path, pipeline.transformer)
wrapper.merge_to()

prompt = "An astronaut is riding a horse through the jungles of Thailand."


## Optional: quantise the model to save on vram.
## Note: The model was quantised during training, and so it is recommended to do the same during inference time.
from optimum.quanto import quantize, freeze, qint8
quantize(pipeline.transformer, weights=qint8)
freeze(pipeline.transformer)
    
pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu') # the pipeline is already in its target precision level
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(42),
    width=1024,
    height=1024,
    guidance_scale=3.0,
).images[0]
image.save("output.png", format="PNG")

Exponential Moving Average (EMA)

SimpleTuner generates a safetensors variant of the EMA weights and a pt file.

The safetensors file is intended to be used for inference, and the pt file is for continuing finetuning.

The EMA model may provide a more well-rounded result, but typically will feel undertrained compared to the full model as it is a running decayed average of the model weights.