From 308ae814d3b73504457ff721fa499263842dbaf5 Mon Sep 17 00:00:00 2001 From: Sayak Paul Date: Mon, 9 Jun 2025 17:42:30 +0530 Subject: [PATCH] Update diffusers-quantization.md --- diffusers-quantization.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/diffusers-quantization.md b/diffusers-quantization.md index 2fff8b47b5..44be461d19 100644 --- a/diffusers-quantization.md +++ b/diffusers-quantization.md @@ -21,7 +21,7 @@ Before we dive into the technical details of how various quantization backends i We created a setup where you can provide a prompt, and we generate results using both the original, high-precision model (e.g., Flux-dev in BF16) and several quantized versions (BnB 4-bit, BnB 8-bit). The generated images are then presented to you and your challenge is to identify which ones came from the quantized models. -Try it out here! +Try it out [here](https://huggingface.co/spaces/diffusers/flux-quant) or below! Often, especially with 8-bit quantization, the differences are subtle and may not be noticeable without close inspection. More aggressive quantization like 4-bit or lower might be more noticeable, but the results can still be good, especially considering the massive memory savings. NF4 often gives the best trade-off though. @@ -592,4 +592,4 @@ Here's a quick guide to choosing a quantization backend: Quantization significantly lowers the barrier to entry for using large diffusion models. Experiment with these backends to find the best balance of memory, speed, and quality for your needs. -*Acknowledgements: Thanks to [Chunte](https://huggingface.co/Chunte) for providing the thumbnail for this post.* \ No newline at end of file +*Acknowledgements: Thanks to [Chunte](https://huggingface.co/Chunte) for providing the thumbnail for this post.*