LCM Dreamshaper v7 — fp16 unet and VAE

Half-precision copies of two components of SimianLuo/LCM_Dreamshaper_v7 (MIT licence), at its commit a85df6a8bd976cdd08b4fd8f3b73f229c9e54df5:

unet/diffusion_pytorch_model.fp16.safetensors
vae/diffusion_pytorch_model.fp16.safetensors

Each tensor of the official diffusion_pytorch_model.safetensors was cast with tensor.to(torch.float16) and saved with safetensors.torch.save_file; nothing else changed. Placed beside the official files they form its fp16 variant for those two components, which the original repository does not publish.

They are used by Image Generator, whose Windows machines without a graphics card hold these weights in fp16 anyway (computing in fp32): fetching them in fp16 halves that part of the download, and the generated image is identical byte for byte. The program checks each file's SHA-256 before using it.

All credit for the model goes to its authors; see the original repository for its card, its limitations and its licence.

Downloads last month: -

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Addey34/LCM_Dreamshaper_v7-fp16

Base model

SimianLuo/LCM_Dreamshaper_v7

Finetuned

(3)

this model