lucid-dl/ddpm-cifar10

DDPM (CIFAR-10 32×32)

https://arxiv.org/abs/2006.11239

Lucid port of diffusers/google/ddpm-cifar10-32, converted to Lucid-native safetensors.

Available weights

Tag	Params	GFLOPs	Size	Source
CIFAR10 (default)	35.7M	—	136.39 MB	diffusers

Usage

import lucid.models as models
from lucid.models.weights import DdpmCifarWeights

# default tag
model = models.ddpm_cifar(pretrained=True)

# explicit tag (enum or string)
model = models.ddpm_cifar(weights=DdpmCifarWeights.CIFAR10)
model = models.ddpm_cifar(pretrained="CIFAR10")

# preprocessing travels with the weights
weights = DdpmCifarWeights.CIFAR10
preprocess = weights.transforms()
out = model(preprocess(image)[None])
logits = out.logits  # (B, num_classes)

Conversion

Converted from diffusers/google/ddpm-cifar10-32 via python -m tools.convert_weights ddpm_cifar --tag CIFAR10. Key mapping + numerical parity verified against the source.

License

apache-2.0 — inherited from the original weights.

Citation

Ho, Jonathan, Ajay Jain, and Pieter Abbeel. "Denoising Diffusion Probabilistic Models." Advances in Neural Information Processing Systems, 2020, pp. 6840–6851.