---
license: apache-2.0
language:
- en
pretty_name: Diffusers Benchmarks
---
Welcome to 🤗 Diffusers Benchmarks!
This is dataset where we keep track of the inference latency and memory information of the core pipelines in the `diffusers` library.
Currently, the core pipelines are the following:
* Stable Diffusion and its derivatives such as ControlNet, T2I Adapter, Image-to-Image, Inpainting
* Stable Diffusion XL and its derivatives
* SSD-1B
* Kandinsky
* Würstchen
* LCM
*Note that we will continue to extend the list of core pipelines based on their API usage.*
We use [this GitHub Actions workflow](https://github.com/huggingface/diffusers/blob/main/.github/workflows/benchmark.yml) to report the above numbers automatically. This workflow runs on a biweekly cadence.
The benchmarks are run on an A10G GPU.