#if !defined(TORCH_STABLE_ONLY) && !defined(TORCH_TARGET_VERSION)
#pragma once

// @generated by torchgen/gen.py from Operator.h

#include <string_view>
#include <tuple>
#include <vector>

// Forward declarations of any types needed in the operator signatures.
// We can't directly include these classes because it will cause circular include dependencies.
// This file is included by TensorBody.h, which defines the Tensor class.
#include <ATen/core/ATen_fwd.h>

namespace at {
namespace _ops {


struct TORCH_API _efficient_attention_forward {
  using schema = ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt> (const at::Tensor &, const at::Tensor &, const at::Tensor &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, const ::std::optional<at::Tensor> &, ::std::optional<c10::SymInt>, ::std::optional<c10::SymInt>, double, int64_t, bool, ::std::optional<double>, const ::std::optional<at::Tensor> &, ::std::optional<int64_t>);
  using ptr_schema = schema*;
  // See Note [static constexpr char* members for windows NVCC]
  static constexpr const char* name = "aten::_efficient_attention_forward";
  static constexpr const char* overload_name = "";
  static constexpr const char* schema_str = "_efficient_attention_forward(Tensor query, Tensor key, Tensor value, Tensor? bias, Tensor? cu_seqlens_q, Tensor? cu_seqlens_k, SymInt? max_seqlen_q, SymInt? max_seqlen_k, float dropout_p, int custom_mask_type, bool compute_log_sumexp=False, *, float? scale=None, Tensor? seqlen_k=None, int? window_size=None) -> (Tensor output, Tensor logsumexp, Tensor philox_seed, Tensor philox_offset, SymInt max_seqlen_batch_q, SymInt max_seqlen_batch_k)";
  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt> call(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, ::std::optional<c10::SymInt> max_seqlen_q, ::std::optional<c10::SymInt> max_seqlen_k, double dropout_p, int64_t custom_mask_type, bool compute_log_sumexp, ::std::optional<double> scale, const ::std::optional<at::Tensor> & seqlen_k, ::std::optional<int64_t> window_size);
  static ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt> redispatch(c10::DispatchKeySet dispatchKeySet, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & bias, const ::std::optional<at::Tensor> & cu_seqlens_q, const ::std::optional<at::Tensor> & cu_seqlens_k, ::std::optional<c10::SymInt> max_seqlen_q, ::std::optional<c10::SymInt> max_seqlen_k, double dropout_p, int64_t custom_mask_type, bool compute_log_sumexp, ::std::optional<double> scale, const ::std::optional<at::Tensor> & seqlen_k, ::std::optional<int64_t> window_size);
};

}} // namespace at::_ops

#else
#error "This file should not be included when either TORCH_STABLE_ONLY or TORCH_TARGET_VERSION is defined."
#endif  // !defined(TORCH_STABLE_ONLY) && !defined(TORCH_TARGET_VERSION)