bakugo-standard-lora-1

This is a standard PEFT LoRA derived from black-forest-labs/FLUX.1-dev.

No validation prompt was used during training.

None

Validation settings

• CFG: 3.5
• 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.

You can find some example images in the following gallery:

Prompt

unconditional (blank prompt)

Negative Prompt

blurry, cropped, ugly

Prompt

A scene from My Hero Academia. Katsuki Bakugo holding a sign that says 'I LOVE PROMPTS!', he is standing full body on a beach at sunset. He is wearing his black and orange hero costume with grenade-like gauntlets on his arms. The setting sun casts a dynamic shadow on his determined expression.

Negative Prompt

blurry, cropped, ugly

Prompt

A scene from My Hero Academia. Katsuki Bakugo jumping out of a propeller airplane, sky diving. He looks intense and exhilarated, his spiky blonde hair blowing in the wind. The sky is clear and blue, with birds flying in the distance.

Negative Prompt

blurry, cropped, ugly

Prompt

A scene from My Hero Academia. Katsuki Bakugo spinning a basketball on his finger on a basketball court. He is wearing a Lakers jersey with the #12 on it. The basketball hoop and crowd are in the background cheering him. He is smirking confidently.

Negative Prompt

blurry, cropped, ugly

Prompt

A scene from My Hero Academia. Katsuki Bakugo is wearing a suit in an office shaking the hand of a businesswoman. The woman has purple hair and is wearing professional attire. There is a Google logo in the background. It is during daytime, and the overall sentiment is one of fiery determination and success.

Negative Prompt

blurry, cropped, ugly

Prompt

A scene from My Hero Academia. Katsuki Bakugo is fighting a large brown grizzly bear, deep in a forest. The bear is tall and standing on two legs, roaring. The bear is also wearing a crown because it is the king of all bears. Around them are tall trees and other animals watching as Bakugo prepares to unleash an explosion.

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: 166

• Training steps: 3000

• Learning rate: 0.0001

  • Learning rate schedule: constant
  • Warmup steps: 100

• Max grad norm: 2.0

• Effective batch size: 48

  • Micro-batch size: 48
  • Gradient accumulation steps: 1
  • Number of GPUs: 1

• Gradient checkpointing: True

• Prediction type: flow-matching (extra parameters=['shift=3', 'flux_guidance_mode=constant', 'flux_guidance_value=1.0', 'flow_matching_loss=compatible', 'flux_lora_target=all'])

• Optimizer: adamw_bf16

• Trainable parameter precision: Pure BF16

• Caption dropout probability: 0.0%

• LoRA Rank: 128

• LoRA Alpha: None

• LoRA Dropout: 0.1

• LoRA initialisation style: default

Datasets

bakugo-512

• Repeats: 2
• Total number of images: 279
• Total number of aspect buckets: 1
• Resolution: 0.262144 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

Inference

import torch
from diffusers import DiffusionPipeline

model_id = 'black-forest-labs/FLUX.1-dev'
adapter_id = 'adipanda/bakugo-standard-lora-1'
pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
pipeline.load_lora_weights(adapter_id)

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=3.5,
).images[0]
image.save("output.png", format="PNG")