Instructions to use Comfy-Org/flux1-kontext-dev_ComfyUI with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusion Single File
How to use Comfy-Org/flux1-kontext-dev_ComfyUI 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
Cap the scale at s <= 448 to avoid saturation?
#4
by Kickbub - opened
Nvidia also caps their scale to <= 448 in transformer engine. Their implementation is a little different because they use amax history.
Here is a comparison:
Seed: 10
Steps: 20
CFG = 1
sampler = euler
scheduler = simple
denoise = 1
prompt = Using this elegant style, create a portrait of a duck wearing a pearl tiara and lace collar, maintaining the same refined quality and soft color tones.
Original image
BF16 weights
Comfy-Org weights
Capped scale weights
The crown stays mostly consistent between the bf16 and capped-scale weights. However this is only 1 comparison
Kickbub changed discussion title from Cap the scale at s <= 448? to Cap the scale at s <= 448 to avoid saturation?



