Update README.md

README.md

@@ -11,19 +11,24 @@ 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. 
+This is dataset where we keep track of the inference latency and memory information of the core models in the `diffusers` library. 
 
-Currently, the core pipelines are the following:
+Currently, the core models are:
 
-* 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
+* Flux
+* Wan
+* LTX
+* SDXL
 
-*Note that we will continue to extend the list of core pipelines based on their API usage.*
+*Note that we will continue to extend this list based on their 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. 
+> [!IMPORTANT]
+> Instead of benchmarking the entire diffusion pipelines, we only benchmark the forward passes
+> of the diffusion networks under different settings (compile, offloading, quantization, etc.).
+> We only benchmark the diffusion networks because they are typically the most compute-heavy part
+> in a diffusion pipeline.
 
-The benchmarks are run on an A10G GPU. 
+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. 
+
+[Here](https://github.com/huggingface/diffusers/actions/runs/16065987231) is an example run.