gravlens-grayscale

This is a full rank finetune derived from kwai-kolors/kolors-diffusers.

The main validation prompt used during training was:

gravitational lensing effects on galaxy

Validation settings

• CFG: 5.0
• CFG Rescale: 0.0
• Steps: 20
• Sampler: None
• Seed: 42
• Resolution: 512x512

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

gravitational lensing effects on galaxy

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: 0
• Training steps: 6750
• Learning rate: 1e-06
  • Learning rate schedule: constant
  • Warmup steps: 675
• Max grad norm: 2.0
• Effective batch size: 8
  • Micro-batch size: 8
  • Gradient accumulation steps: 1
  • Number of GPUs: 1
• Gradient checkpointing: True
• Prediction type: epsilon (extra parameters=['training_scheduler_timestep_spacing=trailing', 'inference_scheduler_timestep_spacing=trailing'])
• Optimizer: optimi-lion
• Trainable parameter precision: Pure BF16
• Caption dropout probability: 10.0%

Datasets

grayscale-lensing-256

• Repeats: 15
• Total number of images: 3689
• Total number of aspect buckets: 1
• Resolution: 0.065536 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

grayscale-lensing-512

• Repeats: 15
• Total number of images: 1801
• 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 = 'GazTrab/gravlens-grayscale'
pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float32) # loading directly in bf16

prompt = "gravitational lensing effects on galaxy"
negative_prompt = 'blurry, cropped, ugly'

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=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=512,
    height=512,
    guidance_scale=5.0,
    guidance_rescale=0.0,
).images[0]
image.save("output.png", format="PNG")