From 9c0944581a386736bc808e68d7dfb52d8cf1790e Mon Sep 17 00:00:00 2001 From: Sayak Paul Date: Fri, 26 Sep 2025 21:50:16 +0530 Subject: [PATCH] [docs] slight edits to the attention backends docs. (#12394) * slight edits to the attention backends docs. * Update docs/source/en/optimization/attention_backends.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> --------- Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> --- docs/source/en/optimization/attention_backends.md | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) diff --git a/docs/source/en/optimization/attention_backends.md b/docs/source/en/optimization/attention_backends.md index 04c8b4ba92..e603878a63 100644 --- a/docs/source/en/optimization/attention_backends.md +++ b/docs/source/en/optimization/attention_backends.md @@ -11,7 +11,7 @@ specific language governing permissions and limitations under the License. --> # Attention backends -> [!TIP] +> [!NOTE] > The attention dispatcher is an experimental feature. Please open an issue if you have any feedback or encounter any problems. Diffusers provides several optimized attention algorithms that are more memory and computationally efficient through it's *attention dispatcher*. The dispatcher acts as a router for managing and switching between different attention implementations and provides a unified interface for interacting with them. @@ -33,7 +33,7 @@ The [`~ModelMixin.set_attention_backend`] method iterates through all the module The example below demonstrates how to enable the `_flash_3_hub` implementation for FlashAttention-3 from the [kernel](https://github.com/huggingface/kernels) library, which allows you to instantly use optimized compute kernels from the Hub without requiring any setup. -> [!TIP] +> [!NOTE] > FlashAttention-3 is not supported for non-Hopper architectures, in which case, use FlashAttention with `set_attention_backend("flash")`. ```py @@ -78,10 +78,16 @@ with attention_backend("_flash_3_hub"): image = pipeline(prompt).images[0] ``` +> [!TIP] +> Most attention backends support `torch.compile` without graph breaks and can be used to further speed up inference. + ## Available backends Refer to the table below for a complete list of available attention backends and their variants. +
+Expand + | Backend Name | Family | Description | |--------------|--------|-------------| | `native` | [PyTorch native](https://docs.pytorch.org/docs/stable/generated/torch.nn.attention.SDPBackend.html#torch.nn.attention.SDPBackend) | Default backend using PyTorch's scaled_dot_product_attention | @@ -104,3 +110,5 @@ Refer to the table below for a complete list of available attention backends and | `_sage_qk_int8_pv_fp16_cuda` | [SageAttention](https://github.com/thu-ml/SageAttention) | INT8 QK + FP16 PV (CUDA) | | `_sage_qk_int8_pv_fp16_triton` | [SageAttention](https://github.com/thu-ml/SageAttention) | INT8 QK + FP16 PV (Triton) | | `xformers` | [xFormers](https://github.com/facebookresearch/xformers) | Memory-efficient attention | + +
\ No newline at end of file