--- license: other base_model: "black-forest-labs/FLUX.1-dev" tags: - flux - flux-diffusers - text-to-image - diffusers - simpletuner - not-for-all-audiences - lora - template:sd-lora - lycoris inference: true widget: - text: 'unconditional (blank prompt)' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_0_0.png - text: 'leirix, a woman with blonde hair in pigtails with pink and blue highlights, wearing a sleeveless black and red crop top and matching high waisted pants, with long red gloves, holding a large yellow fish over her shoulder, with a playful and happy expression, looking to the side, wearing bold red lipstick, with a small heart tattoo under her eye, background features a red backdrop with white stars' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_1_0.png - text: 'leirix, a young woman enjoying a sunny day at the beach, with long, flowing blonde hair that catches the light beautifully and bright, expressive blue eyes, wearing a sleek, one-piece athletic swimsuit with shades of blue and white, perfectly complementing her sun-kissed skin, sitting on a light blue beach towel, her posture relaxed as she leans back slightly, with the serene ocean and sandy beach stretching out behind her, sunlight casts a warm glow, creating soft shadows and highlighting the contours of her figure, scene exudes a peaceful, summery vibe, with the clear sky and calm waves in the background' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_2_0.png - text: 'leirix, a woman with long wavy dark hair, wearing a strapless form-fitting red dress with a sweetheart neckline and matching long gloves, her expression is confident and alluring, she is adorned with dangling earrings, she has one arm raised above her head, the background is a solid pink color' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_3_0.png - text: 'leirix, a woman with long wavy dark hair, wearing a red bikini, she has a red flower in her hair and is adorned with pearl earrings, her expression is confident and alluring, her body is glistening with water as she stands on a beach, the background features a clear blue sky, palm leaves, and the ocean' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_4_0.png - text: 'leirix, a woman with red hair, wearing a silver chainmail top and a large fur cloak, she is adorned with gold earrings and a necklace, she is riding a brown horse and holding the reins tightly, the background is a snowy landscape with a clear sky' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_5_0.png - text: 'leirix, a woman with vibrant red hair styled in an elaborate voluminous fashion, she has green, glowing eyes, wearing a form-fitting green outfit adorned with various plant elements and red berries, her hand is raised as if casting a spell, the background is blurred with soft, muted tones of green and blue' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_6_0.png --- # leirix-flux-lokr-lion-1e-5-bs4-ga6-v03 This is a LyCORIS adapter derived from [black-forest-labs/FLUX.1-dev](https://huggingface.co/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](#training-settings). You can find some example images in the following gallery: The text encoder **was not** trained. You may reuse the base model text encoder for inference. ## Training settings - Training epochs: 57 - Training steps: 8840 - Learning rate: 1e-05 - Learning rate schedule: polynomial - Warmup steps: 100 - Max grad norm: 0.01 - Effective batch size: 12 - Micro-batch size: 2 - Gradient accumulation steps: 3 - Number of GPUs: 2 - Gradient checkpointing: True - Prediction type: flow-matching (extra parameters=['flux_schedule_auto_shift', 'shift=0.0', 'flux_guidance_mode=constant', 'flux_guidance_value=1.0', 'flow_matching_loss=compatible']) - Optimizer: optimi-lionweight_decay=1e-3 - Trainable parameter precision: Pure BF16 - Caption dropout probability: 10.0% ### LyCORIS Config: ```json { "algo": "lokr", "multiplier": 1.0, "linear_dim": 10000, "linear_alpha": 1, "factor": 12, "apply_preset": { "target_module": [ "Attention", "FeedForward" ], "module_algo_map": { "Attention": { "factor": 12 }, "FeedForward": { "factor": 6 } } } } ``` ## Datasets ### LEIRIX-FLUX-LLM-V03-512 - Repeats: 1 - Total number of images: ~146 - Total number of aspect buckets: 2 - Resolution: 0.262144 megapixels - Cropped: True - Crop style: random - Crop aspect: closest - Used for regularisation data: No ### LEIRIX-FLUX-LLM-V03-768 - Repeats: 1 - Total number of images: ~148 - Total number of aspect buckets: 3 - Resolution: 0.589824 megapixels - Cropped: True - Crop style: random - Crop aspect: closest - Used for regularisation data: No ### LEIRIX-FLUX-LLM-V03-1024 - Repeats: 1 - Total number of images: ~148 - Total number of aspect buckets: 1 - Resolution: 1.048576 megapixels - Cropped: True - Crop style: random - Crop aspect: closest - Used for regularisation data: No ### LEIRIX-FLUX-WD-V03-512 - Repeats: 1 - Total number of images: ~146 - Total number of aspect buckets: 1 - Resolution: 0.262144 megapixels - Cropped: True - Crop style: random - Crop aspect: closest - Used for regularisation data: No ### LEIRIX-FLUX-WD-V03-768 - Repeats: 1 - Total number of images: ~148 - Total number of aspect buckets: 1 - Resolution: 0.589824 megapixels - Cropped: True - Crop style: random - Crop aspect: closest - Used for regularisation data: No ### LEIRIX-FLUX-WD-V03-1024 - Repeats: 1 - Total number of images: ~148 - Total number of aspect buckets: 4 - Resolution: 1.048576 megapixels - Cropped: True - Crop style: random - Crop aspect: closest - Used for regularisation data: No ## Inference ```python 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 = 'gattaplayer/leirix-flux-lokr-lion-1e-5-bs4-ga6-v03' 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") ```