SimpleTuner/flux-3-klein-image-training-assistant

This is a PEFT LoRA derived from black-forest-labs/flux-3-image-klein.

No validation prompt was used during training.

Validation settings

• CFG: 1.0
• CFG Rescale: 0.0
• Steps: 16
• Sampler: FlowMatchEulerDiscreteScheduler
• Seed: 42
• Resolutions: 512x512,1024x1024,1536x1536,2048x2048

Note: The validation settings are not necessarily the same as the training settings.

You can find some example images in the following gallery:

Prompt

A photograph of an elderly woman wearing a blue silk scarf, soft window light, a plain gray background.

Negative Prompt

'

Prompt

A photograph of an elderly woman wearing a blue silk scarf, soft window light, a plain gray background.

Negative Prompt

'

Prompt

A photograph of an elderly woman wearing a blue silk scarf, soft window light, a plain gray background.

Negative Prompt

'

Prompt

A photograph of an elderly woman wearing a blue silk scarf, soft window light, a plain gray background.

Negative Prompt

'

Prompt

A photograph of a woman walking down a rain-soaked city street, reflections in the pavement, natural evening light.

Negative Prompt

'

Prompt

A photograph of a woman walking down a rain-soaked city street, reflections in the pavement, natural evening light.

Negative Prompt

'

Prompt

A photograph of a woman walking down a rain-soaked city street, reflections in the pavement, natural evening light.

Negative Prompt

'

Prompt

A photograph of a woman walking down a rain-soaked city street, reflections in the pavement, natural evening light.

Negative Prompt

'

Prompt

A photograph of a sunlit kitchen with wooden cabinets, ceramic cups and fresh flowers on a table.

Negative Prompt

'

Prompt

A photograph of a sunlit kitchen with wooden cabinets, ceramic cups and fresh flowers on a table.

Negative Prompt

'

Prompt

A photograph of a sunlit kitchen with wooden cabinets, ceramic cups and fresh flowers on a table.

Negative Prompt

'

Prompt

A photograph of a sunlit kitchen with wooden cabinets, ceramic cups and fresh flowers on a table.

Negative Prompt

'

Prompt

A photograph of a mountain lake at dawn with mist over the water and pine trees along the shore.

Negative Prompt

'

Prompt

A photograph of a mountain lake at dawn with mist over the water and pine trees along the shore.

Negative Prompt

'

Prompt

A photograph of a mountain lake at dawn with mist over the water and pine trees along the shore.

Negative Prompt

'

Prompt

A photograph of a mountain lake at dawn with mist over the water and pine trees along the shore.

Negative Prompt

'

Prompt

A close-up photograph of a white cotton shirt and a folded blue silk scarf on a wooden table.

Negative Prompt

'

Prompt

A close-up photograph of a white cotton shirt and a folded blue silk scarf on a wooden table.

Negative Prompt

'

Prompt

A close-up photograph of a white cotton shirt and a folded blue silk scarf on a wooden table.

Negative Prompt

'

Prompt

A close-up photograph of a white cotton shirt and a folded blue silk scarf on a wooden table.

Negative Prompt

'

Prompt

An illustration of an anthropomorphic fox wearing a blue jacket and reading a book in a cozy library.

Negative Prompt

'

Prompt

An illustration of an anthropomorphic fox wearing a blue jacket and reading a book in a cozy library.

Negative Prompt

'

Prompt

An illustration of an anthropomorphic fox wearing a blue jacket and reading a book in a cozy library.

Negative Prompt

'

Prompt

An illustration of an anthropomorphic fox wearing a blue jacket and reading a book in a cozy library.

Negative Prompt

'

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

Training settings

• Training epochs: 0

• Training steps: 30000

• Learning rate: 0.0001

  • Learning rate schedule: constant_with_warmup
  • Warmup steps: 25

• Max grad norm: 1.0

• Effective batch size: 16

  • Micro-batch size: 2
  • Gradient accumulation steps: 1
  • Number of GPUs: 8

• Gradient checkpointing: True

• Prediction type: flow_map[]

• Optimizer: adamw_bf16

• Trainable parameter precision: Pure BF16

• Base model precision: no_change

• Caption dropout probability: 0.0%

• LoRA Rank: 64

• LoRA Alpha: 64.0

• LoRA Dropout: 0.0

• LoRA initialisation style: default

• LoRA mode: Standard

Datasets

assistant-real-v3-cc12m-512

• Repeats: 0
• Total number of images: 15000
• Total number of aspect buckets: 56
• Resolution: 0.262144 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

assistant-real-v3-cc12m-1024

• Repeats: 0
• Total number of images: 15000
• Total number of aspect buckets: 96
• Resolution: 1.048576 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

assistant-real-v3-cc12m-1536

• Repeats: 0
• Total number of images: 15000
• Total number of aspect buckets: 127
• Resolution: 2.359296 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

assistant-real-v3-cc12m-2048

• Repeats: 0
• Total number of images: 15000
• Total number of aspect buckets: 153
• Resolution: 4.194304 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

assistant-real-v3-e621-512

• Repeats: 0
• Total number of images: 15000
• Total number of aspect buckets: 86
• Resolution: 0.262144 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

assistant-real-v3-e621-1024

• Repeats: 0
• Total number of images: 15000
• Total number of aspect buckets: 142
• Resolution: 1.048576 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

assistant-real-v3-e621-1536

• Repeats: 0
• Total number of images: 15000
• Total number of aspect buckets: 177
• Resolution: 2.359296 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

assistant-real-v3-e621-2048

• Repeats: 0
• Total number of images: 15000
• Total number of aspect buckets: 201
• Resolution: 4.194304 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

Inference

import torch
from simpletuner.helpers.models.flux3_image.pipeline import Flux3ImagePipeline as DiffusionPipeline

model_id = 'black-forest-labs/flux-3-image-klein'
adapter_id = 'SimpleTuner/flux-3-klein-image-training-assistant'
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."
negative_prompt = ''

## 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
model_output = pipeline(
    prompt=prompt,
    negative_prompt=negative_prompt,
    num_inference_steps=16,
    generator=torch.Generator(device='cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu').manual_seed(42),
    width=512,
    height=512,
    guidance_scale=1.0,
).images[0]

model_output.save("output.png", format="PNG")