FLUX.2-DEV-DLAY-LyCORIS

FLUX.2-DEV-DLAY-LyCORIS

This is a LyCORIS adapter derived from black-forest-labs/FLUX.2-dev.

No validation prompt was used during training.

None

Validation settings

• CFG: 4.0
• CFG Rescale: 0.0
• Steps: 28
• Sampler: FlowMatchEulerDiscreteScheduler
• 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:

<Gallery />

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

Training settings

• Training epochs: 8
• Training steps: 25000
• Learning rate: 1.0
  • Learning rate schedule: constant
  • Warmup steps: 0
• Max grad norm: 1.0
• Effective batch size: 8
  • Micro-batch size: 1
  • Gradient accumulation steps: 1
  • Number of GPUs: 8
• Gradient checkpointing: False
• Prediction type: flow_matching[]
• Optimizer: prodigy (config=d_coef=2.0,weight_decay=0.01,use_bias_correction=False,schedulefree_c=75,use_orthograd=True)
• Trainable parameter precision: Pure BF16
• Base model precision: no_change
• Caption dropout probability: 0.1%

LyCORIS Config:

{
    "algo": "lokr",
    "multiplier": 1.0,
    "factor": 12,
    "linear_dim": 100000,
    "linear_alpha": 1,
    "conv_dim": 0,
    "conv_alpha": 0,
    "dora_wd": true,
    "apply_preset": {
        "target_module": [
            "Flux2TransformerBlock",
            "Flux2SingleTransformerBlock"
        ]
    }
}

Datasets

dlay-subject-data

• Repeats: 10
• Total number of images: ~2152
• Total number of aspect buckets: 1
• Resolution: 1.048576 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• 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.2-dev'
adapter_repo_id = 'ThatDustyGuy/FLUX.2-DEV-DLAY-LyCORIS'
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."
negative_prompt = 'blurry, cropped, ugly'

## Optional: quantise the model to save on vram.
## Note: The model was not quantised during training, so it is not necessary to quantise it 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
model_output = pipeline(
    prompt=prompt,
    negative_prompt=negative_prompt,
    num_inference_steps=28,
    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=4.0,
).images[0]

model_output.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.