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
Metadata lost from model
#1
by mcmonkey - opened
My bad, I didn't actually even know that's a thing. Should be fixed now (if I did it right). Thanks for letting me know.
Do I leave the weight_dtype setting on default or should I set it to fp8 when using this model?
Edit: Thank you for the upload
Do I leave the weight_dtype setting on default or should I set it to fp8 when using this model?
On default it loads it as whatever is your Comfy default, which is probably bf or fp16, so you do have to change it to fp8_e4m3fnon the node even when using this model.
Thank you so much for uploading the model. Hopefully 4 steps model in the future
