growwithdaisy/vnlwnxgstfwstmn_style_flat_20241120_174945

This is a LyCORIS adapter derived from FLUX.1-dev.

The main validation prompt used during training was:

a photo of a daisy

Validation settings

• CFG: 3.5
• CFG Rescale: 0.0
• Steps: 20
• Sampler: FlowMatchEulerDiscreteScheduler
• Seed: 69
• 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

mini white gstfwstmn chunky candle holder with a candle inside, white background, jmmymrbl cutup photography style

Negative Prompt

blurry, cropped, ugly

Prompt

vnlwn ice cream, in a cup or tub, three scoops, malted milkshake and fries flavor, white background, jmmymrbl dune photography style

Negative Prompt

blurry, cropped, ugly

Prompt

three children sitting on a door stoop, eating candy, wearing winter clothes, holding a bag, wooden door in background, boy, pttknt blonde child, infant, jmmymrbl earthly delights photography style

Negative Prompt

blurry, cropped, ugly

Prompt

gstfwstmn chunky desk, furry white chair under desk, blue vase on top, mirrors on the walls, next to a strange small table, ktyshyn photography style

Negative Prompt

blurry, cropped, ugly

Prompt

a photo of a daisy

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: 132
• Training steps: 15000
• Learning rate: 0.0001
  • Learning rate schedule: constant
  • Warmup steps: 0
• Max grad norm: 2.0
• Effective batch size: 16
  • Micro-batch size: 2
  • Gradient accumulation steps: 1
  • Number of GPUs: 8
• 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: optimi-stableadamwweight_decay=1e-3
• Trainable parameter precision: Pure BF16
• Caption dropout probability: 5.0%

LyCORIS Config:

{
    "algo": "lokr",
    "multiplier": 1,
    "linear_dim": 1000000,
    "linear_alpha": 1,
    "factor": 12,
    "init_lokr_norm": 0.001,
    "apply_preset": {
        "target_module": [
            "FluxTransformerBlock",
            "FluxSingleTransformerBlock"
        ],
        "module_algo_map": {
            "Attention": {
                "factor": 12
            },
            "FeedForward": {
                "factor": 6
            }
        }
    }
}

Datasets

vnlwnxgstfwstmn_style_flat-512

• Repeats: 0
• Total number of images: ~296
• Total number of aspect buckets: 1
• Resolution: 0.262144 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

vnlwnxgstfwstmn_style_flat-768

• Repeats: 0
• Total number of images: ~248
• Total number of aspect buckets: 1
• Resolution: 0.589824 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

vnlwnxgstfwstmn_style_flat-1024

• Repeats: 1
• Total number of images: ~184
• Total number of aspect buckets: 7
• 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 = 'FLUX.1-dev'
adapter_repo_id = 'playerzer0x/growwithdaisy/vnlwnxgstfwstmn_style_flat_20241120_174945'
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 = "a photo of a daisy"


## 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
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(69),
    width=1024,
    height=1024,
    guidance_scale=3.5,
).images[0]
image.save("output.png", format="PNG")