easydel.kernels.gpu_ops.flash_attention_triton._flash_attention
easydel.kernels.gpu_ops.flash_attention_triton._flash_attention
-
easydel.kernels.gpu_ops.flash_attention_triton._flash_attention.flash_attention(q: Optional[Union[Array, ndarray, bool, number]], k: Optional[Union[Array, ndarray, bool, number]], v: Optional[Union[Array, ndarray, bool, number]], attention_mask: Optional[Union[Array, ndarray, bool, number]] = None, bias: Optional[Union[Array, ndarray, bool, number]] = None, softmax_scale: Optional[float] = None, dropout_prob: float = 0.0, causal: bool = False, dropout_seed: Optional[int] = None, varlen_mode: bool = True) → Union[Array, ndarray, bool, number][source]