Hugging Face's logo

bghira
/
sd35m-photo-SLG-shift0.8-attn_ff-512to2048px

Text-to-Image
Diffusers
sd3
sd3-diffusers
simpletuner
Not-For-All-Audiences
lora
template:sd-lora
standard
Model card Files
xet
Community

sd35m-photo-SLG-shift0.8-attn_ff-512to2048px

This is a standard PEFT LoRA derived from stabilityai/stable-diffusion-3.5-medium.

The main validation prompt used during training was:

A photo-realistic image of a cat

Validation settings

  • CFG: 6.0
  • CFG Rescale: 0.0
  • Steps: 30
  • Sampler: FlowMatchEulerDiscreteScheduler
  • Seed: 42
  • Resolution: 1024x1024
  • Skip-layer guidance: skip_guidance_layers=[7, 8, 9],

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
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Alien planet, strange rock formations, glowing plants, bizarre creatures, surreal atmosphere
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Alien marketplace, bizarre creatures, exotic goods, vibrant colors, otherworldly atmosphere
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Child holding a balloon, happy expression, colorful balloons, sunny day, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
a 4-panel comic strip showing an orange cat saying the words 'HELP' and 'LASAGNA'
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
a hand is holding a comic book with a cover that reads 'The Adventures of Superhero'
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Underground cave filled with crystals, glowing lights, reflective surfaces, fantasy environment, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Bustling cyberpunk bazaar, vendors, neon signs, advanced tech, crowded, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Cyberpunk hacker in a dark room, neon glow, multiple screens, intense focus, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
a cybernetic anne of green gables with neural implant and bio mech augmentations
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Post-apocalyptic cityscape, ruined buildings, overgrown vegetation, dark and gritty, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Magical castle in a lush forest, glowing windows, fantasy architecture, high resolution, detailed textures
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Ruins of an ancient temple in an enchanted forest, glowing runes, mystical creatures, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Mystical forest, glowing plants, fairies, magical creatures, fantasy art, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Magical garden with glowing flowers, fairies, serene atmosphere, detailed plants, high resolution
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Whimsical garden filled with fairies, magical plants, sparkling lights, serene atmosphere, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Majestic dragon soaring through the sky, detailed scales, dynamic pose, fantasy art, high resolution
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Fantasy world, floating islands in the sky, waterfalls, lush vegetation, detailed landscape, high resolution
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Futuristic city skyline at night, neon lights, cyberpunk style, high contrast, sharp focus
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Space battle scene, starships fighting, laser beams, explosions, cosmic background
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Abandoned fairground at night, eerie rides, ghostly figures, fog, dark atmosphere, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Spooky haunted mansion on a hill, dark and eerie, glowing windows, ghostly atmosphere, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
a hardcover physics textbook that is called PHYSICS FOR DUMMIES
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Epic medieval battle, knights in armor, dynamic action, detailed landscape, high resolution
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Bustling medieval market with merchants, knights, and jesters, vibrant colors, detailed
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Cozy medieval tavern, warm firelight, adventurers drinking, detailed interior, rustic atmosphere
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Futuristic city skyline at night, neon lights, cyberpunk style, high contrast, sharp focus
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Forest with neon-lit trees, glowing plants, bioluminescence, surreal atmosphere, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Bright neon sign in a busy city street, 'Open 24 Hours', bold typography, glowing lights
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Vibrant neon sign, 'Bar', bold typography, dark background, glowing lights, detailed design
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Pirate ship on the high seas, stormy weather, detailed sails, dramatic waves, photorealistic
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Pirate discovering a treasure chest, detailed gold coins, tropical island, dramatic lighting
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
a photograph of a woman experiencing a psychedelic trip. trippy, 8k, uhd, fractal
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Cozy cafe on a rainy day, people sipping coffee, warm lights, reflections on wet pavement, photorealistic
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
1980s arcade, neon lights, vintage game machines, kids playing, vibrant colors, nostalgic atmosphere
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
1980s game room with vintage arcade machines, neon lights, vibrant colors, nostalgic feel
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Robot blacksmith forging metal, sparks flying, detailed workshop, futuristic and medieval blend
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Sleek robot performing a dance, futuristic theater, holographic effects, detailed, high resolution
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
High-tech factory where robots are assembled, detailed machinery, futuristic setting, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Garden tended by robots, mechanical plants, colorful flowers, futuristic setting, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Cute robotic pet, futuristic home, sleek design, detailed features, friendly and animated
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
cctv trail camera night time security picture of a wendigo in the woods
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Astronaut exploring an alien planet, detailed landscape, futuristic suit, cosmic background
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Futuristic space station orbiting a distant exoplanet, sleek design, detailed structures, cosmic backdrop
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
a person holding a sign that reads 'SOON'
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Steampunk airship in the sky, intricate design, Victorian aesthetics, dynamic scene, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Steampunk inventor in a workshop, intricate gadgets, Victorian attire, mechanical arm, goggles
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Stormy ocean with towering waves, dramatic skies, detailed water, intense atmosphere, high resolution
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Dramatic stormy sea, lighthouse in the distance, lightning striking, dark clouds, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Graffiti artist creating a mural, vibrant colors, urban setting, dynamic action, high resolution
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Urban alleyway filled with vibrant graffiti art, tags and murals, realistic textures
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Urban street sign, 'Main Street', bold typography, realistic textures, weathered look
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Classic car show with vintage vehicles, vibrant colors, nostalgic atmosphere, high detail
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Retro diner sign, 'Joe's Diner', classic 1950s design, neon lights, weathered look
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
Vintage store sign with elaborate typography, 'Antique Shop', hand-painted, weathered look
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average
Prompt
A photo-realistic image of a cat
Negative Prompt
ugly, cropped, blurry, low-quality, mediocre average

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

Training settings

  • Training epochs: 1

  • Training steps: 64000

  • Learning rate: 0.0003

    • Learning rate schedule: constant
    • Warmup steps: 500

  • Max grad norm: 0.01

  • Effective batch size: 9

    • Micro-batch size: 3
    • Gradient accumulation steps: 1
    • Number of GPUs: 3

  • Gradient checkpointing: True

  • Prediction type: flow-matching (extra parameters=['flux_schedule_auto_shift', 'shift=0.0', 'flux_use_uniform_schedule'])

  • Optimizer: adamw_bf16

  • Trainable parameter precision: Pure BF16

  • Caption dropout probability: 25.0%

  • LoRA Rank: 128

  • LoRA Alpha: 128.0

  • LoRA Dropout: 0.1

  • LoRA initialisation style: default

Datasets

signs

  • Repeats: 0
  • Total number of images: ~420
  • Total number of aspect buckets: 14
  • Resolution: 1.048576 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

moviecollection

  • Repeats: 0
  • Total number of images: ~1983
  • Total number of aspect buckets: 34
  • Resolution: 1.048576 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

bookcovers

  • Repeats: 0
  • Total number of images: ~927
  • Total number of aspect buckets: 26
  • Resolution: 1.048576 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

shutterstock

  • Repeats: 0
  • Total number of images: ~21111
  • Total number of aspect buckets: 33
  • Resolution: 1.048576 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

cinemamix-1mp

  • Repeats: 0
  • Total number of images: ~7425
  • Total number of aspect buckets: 5
  • Resolution: 1.048576 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

anatomy

  • Repeats: 5
  • Total number of images: ~16440
  • Total number of aspect buckets: 24
  • Resolution: 1.048576 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

signs-512

  • Repeats: 0
  • Total number of images: ~417
  • Total number of aspect buckets: 12
  • Resolution: 0.262144 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

moviecollection-512

  • Repeats: 0
  • Total number of images: ~1971
  • Total number of aspect buckets: 19
  • Resolution: 0.262144 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

bookcovers-512

  • Repeats: 0
  • Total number of images: ~918
  • Total number of aspect buckets: 16
  • Resolution: 0.262144 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

shutterstock-512

  • Repeats: 0
  • Total number of images: ~21096
  • Total number of aspect buckets: 19
  • Resolution: 0.262144 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

cinemamix-1mp-512

  • Repeats: 0
  • Total number of images: ~7422
  • Total number of aspect buckets: 3
  • Resolution: 0.262144 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

anatomy-512

  • Repeats: 5
  • Total number of images: ~16437
  • Total number of aspect buckets: 13
  • Resolution: 0.262144 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

signs-1440

  • Repeats: 100
  • Total number of images: ~423
  • Total number of aspect buckets: 1
  • Resolution: 2.0736 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

moviecollection-1440

  • Repeats: 0
  • Total number of images: ~2007
  • Total number of aspect buckets: 44
  • Resolution: 2.0736 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

bookcovers-1440

  • Repeats: 0
  • Total number of images: ~933
  • Total number of aspect buckets: 29
  • Resolution: 2.0736 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

shutterstock-1440

  • Repeats: 0
  • Total number of images: ~21111
  • Total number of aspect buckets: 42
  • Resolution: 2.0736 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

cinemamix-1mp-1440

  • Repeats: 0
  • Total number of images: ~7425
  • Total number of aspect buckets: 5
  • Resolution: 2.0736 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

anatomy-1440

  • Repeats: 5
  • Total number of images: ~16458
  • Total number of aspect buckets: 32
  • Resolution: 2.0736 megapixels
  • Cropped: False
  • Crop style: None
  • Crop aspect: None
  • Used for regularisation data: No

Inference 

import torch
from diffusers import DiffusionPipeline

model_id = 'stabilityai/stable-diffusion-3.5-medium'
adapter_id = 'bghira/sd35m-photo-SLG-shift0.8-attn_ff-512to2048px'
pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
pipeline.load_lora_weights(adapter_id)

prompt = "A photo-realistic image of a cat"
negative_prompt = 'ugly, cropped, blurry, low-quality, mediocre average'

## 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,
    negative_prompt=negative_prompt,
    num_inference_steps=30,
    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=6.0,
    skip_guidance_layers=[7, 8, 9],
).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.

Downloads last month
0
Inference Examples
Text-to-Image
Examples
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for bghira/sd35m-photo-SLG-shift0.8-attn_ff-512to2048px

Base model

stabilityai/stable-diffusion-3.5-medium
Adapter
(71)
this model