lucid-dl/stable-diffusion-v1
0
High-Resolution Image Synthesis with Latent Diffusion Models
https://arxiv.org/abs/2112.10752
Lucid port of https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5, converted to Lucid-native safetensors.
Available weights
Usage
import lucid
import lucid.models as models
from lucid.models.weights import Stable diffusionWeights
# default tag
model = models.stable_diffusion_v1(pretrained=True)
# explicit tag (enum or string)
model = models.stable_diffusion_v1(weights=Stable diffusionWeights.CompVis_LAION)
model = models.stable_diffusion_v1(pretrained="CompVis_LAION")
# feed token ids (tokenize with the matching lucid.utils.tokenizer)
input_ids = lucid.tensor([[101, 7592, 2088, 102]], dtype=lucid.int64)
out = model(input_ids)
hidden = out.last_hidden_state # (B, T, hidden_size)Conversion
Converted from https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5 via python -m tools.convert_weights Stable Diffusion --tag CompVis_LAION. Key mapping + numerical parity verified against the source.
License
creativeml-openrail-m — inherited from the original weights.
Citation
@inproceedings{rombach2022high,
title={High-Resolution Image Synthesis with Latent Diffusion Models},
author={Rombach, Robin and Blattmann, Andreas and Lorenz, Dominik and Esser, Patrick and Ommer, Bj{\"o}rn},
booktitle={CVPR},
pages={10684--10695},
year={2022}
}