Instructions to use Kijai/flux-fp8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusion Single File
How to use Kijai/flux-fp8 with Diffusion Single File:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Is your dev-fp8-e4m3fn version consistent with the official fp8 version released by Comfyui?
#33
by HUG-NAN - opened
comfyui : https://huggingface.co/Comfy-Org/flux1-dev/tree/main
Your version is 11G in size and only quantifies the UNet part, while the version provided by Comfyui is 17G in size and includes UNet, clip, VAE
You kinda answered the question already? That is the only difference, yes.
You kinda answered the question already? That is the only difference, yes.
Has the 17g version of Comfyui fully quantified UNet+clip+VAE? Are the running speed and quality of your two versions consistent?
When it comes to the diffusion model, as far as I know there is no difference.