Skip to content

vllm_omni.diffusion.attention.backends.flash_attn

logger module-attribute 

logger = init_logger(__name__)

FlashAttentionBackend

Bases: AttentionBackend

accept_output_buffer class-attribute instance-attribute 

accept_output_buffer: bool = True

get_impl_cls staticmethod 

get_impl_cls() -> type[FlashAttentionImpl]

get_name staticmethod 

get_name() -> str

get_supported_head_sizes staticmethod 

get_supported_head_sizes() -> list[int]

supports_attention_mask classmethod 

supports_attention_mask() -> bool

FlashAttentionImpl

Bases: AttentionImpl

causal instance-attribute 

causal = causal

num_heads instance-attribute 

num_heads = num_heads

qkv_layout instance-attribute 

qkv_layout = qkv_layout

softmax_scale instance-attribute 

softmax_scale = softmax_scale

forward_cuda 

forward_cuda(
    query: Tensor,
    key: Tensor,
    value: Tensor,
    attn_metadata: AttentionMetadata = None,
) -> Tensor

CUDA/ROCm/MUSA flash attention implementation.

forward_fa_npu 

forward_fa_npu(
    query: Tensor,
    key: Tensor,
    value: Tensor,
    attn_metadata: AttentionMetadata = None,
) -> Tensor

forward_fa_quant_npu 

forward_fa_quant_npu(
    query: Tensor,
    key: Tensor,
    value: Tensor,
    attn_metadata: AttentionMetadata = None,
) -> Tensor

forward_npu 

forward_npu(
    query: Tensor,
    key: Tensor,
    value: Tensor,
    attn_metadata: AttentionMetadata = None,
) -> Tensor

NPU attention implementation using mindiesd.

forward_xpu 

forward_xpu(
    query: Tensor,
    key: Tensor,
    value: Tensor,
    attn_metadata: AttentionMetadata = None,
) -> Tensor

XPU flash attention implementation.