Hugging Face's logo

playerzer0x
/
y

Text-to-Image
Diffusers
flux
flux-diffusers
simpletuner
Not-For-All-Audiences
lora
template:sd-lora
lycoris
Model card Files
xet
 Community

y

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
bssprshps car interior objects, black background, vertically positioned, antenna
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps car interior objects, car dashboard cover, gray, car interior background, on top of the dashboard, side view
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps car interior objects, windshield phone mount, on top of a silky cloth, side view, silky cloth background
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps car interior objects, windshield phone mount, car interior background, sticking on the windshield, isometric view
Negative Prompt
blurry, cropped, ugly
Prompt
snwypnswhtlbs dog, front view, sitting position, tongue is out, collar on its neck and leash is lifted up, grass field on the back, sitting on a cement ground
Negative Prompt
blurry, cropped, ugly
Prompt
snwypnswhtlbs dog, front view, out-of-focus background, laying on a white blanket, red ribbon on its neck
Negative Prompt
blurry, cropped, ugly
Prompt
snwypnswhtlbs dog, laying on the water, tongue is out, out-of-focus background, pond background, facing on the right side
Negative Prompt
blurry, cropped, ugly
Prompt
snwypnswhtlbs dog, running in the water, out-of-focus background, lake background, running to the left, side view
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps cooler, brown and black, white background, front view
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps cooler, opened, red, front view, garden background, on the ground, filled with drinks and fruits and bread
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps cooler, front view, boat on water background, on top of a boat, light skinned person closing the box
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps cooler, turquoise, opened, isometric view, white background
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq perfume, white background, next to a red coral and oval glass, corail oscuro scent, product is laying
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq perfume, white background, product is laying, bottle cap is next to the product, top view, doson scent, perfume bottle is open, oval bottle and cylinder-shaped cap
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq perfume, oval bottle and cylinder-shaped cap, product is laying, leau papier scent, on the floor, next to a writing materials
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq perfume, oval bottle and cylinder-shaped cap, leau papier scent, product is laying, top view, next to the product branding
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps hat, orange, white background, isometric view, corduroy
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps hat, brown, on top of the couch, front view, navy blue couch, text only logo
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps hat, sky blue, black background, isometric view
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps hat, camo brown and green, black background, isometric view
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq perfume set, white background, cylinder-shaped bottles, front view
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq perfume set, white background, cylinder-shaped bottles, front view
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq perfume set, front view, on the packaging box, next to the packaging cover, black background
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq perfume set, white background, front view, on the packaging box, next to the packaging cover
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq car diffuser, white background, amber scent, back view, next to the product packaging
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq car diffuser, white background, roses scent, back view, next to the product packaging
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq car diffuser, white background, back view
Negative Prompt
blurry, cropped, ugly
Prompt
dptyq car diffuser, white background, back view, ginger scent, next to the product packaging
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps head net, worn by a tan-skinned woman, worn by a dark-skinned man, out-of-focus background, front view, taking a selfie, wearing a camo hat and a teal and gray striped shirt, wearing a woven hat and a flannel shirt
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps head net, worn by a mannequin head, side view, white background, wearing a white hat
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps head net, worn by a light-skinned man, side view, white background, wearing a black baseball cap and a white shirt
Negative Prompt
blurry, cropped, ugly
Prompt
bssprshps head net, out-of-focus background, worn by a light-skinned woman, side view, green, wearing a knitted hat, glasses, and a dark shirt
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: 40
  • Training steps: 4000
  • Learning rate: 0.0001
    • Learning rate schedule: constant
    • Warmup steps: 0
  • Max grad norm: 2.0
  • Effective batch size: 8
    • Micro-batch size: 2
    • Gradient accumulation steps: 1
    • Number of GPUs: 4
  • 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

dptyqxbssprshps_flat-512

  • Repeats: 0
  • Total number of images: ~216
  • Total number of aspect buckets: 1
  • Resolution: 0.262144 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

dptyqxbssprshps_flat-768

  • Repeats: 0
  • Total number of images: ~180
  • Total number of aspect buckets: 2
  • Resolution: 0.589824 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

dptyqxbssprshps_flat-1024

  • Repeats: 1
  • Total number of images: ~152
  • Total number of aspect buckets: 11
  • 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/y'
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")
Downloads last month
8
Inference Examples
Text-to-Image
Examples
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support