@@ -29,7 +29,7 @@ However, it is hard to decide when to reuse the cache to ensure quality generate
...
@@ -29,7 +29,7 @@ However, it is hard to decide when to reuse the cache to ensure quality generate
This achieves a 2x speedup on FLUX.1-dev and HunyuanVideo inference with very good quality.
This achieves a 2x speedup on FLUX.1-dev and HunyuanVideo inference with very good quality.
<figure>
<figure>
<imgsrc="https://huggingface.co/datasets/chengzeyi/documentation-images/resolve/main/diffusers/para-attn/ada-cache.png"alt="Cache in Diffusion Transformer"/>
<imgsrc="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/para-attn/ada-cache.png"alt="Cache in Diffusion Transformer"/>
<figcaption>How AdaCache works, First Block Cache is a variant of it</figcaption>
<figcaption>How AdaCache works, First Block Cache is a variant of it</figcaption>