#pragma once

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

#include <ATen/Context.h>
#include <ATen/DeviceGuard.h>
#include <ATen/TensorUtils.h>
#include <ATen/TracerMode.h>
#include <ATen/core/Generator.h>
#include <ATen/core/Reduction.h>
#include <ATen/core/Tensor.h>
#include <c10/core/Scalar.h>
#include <c10/core/Storage.h>
#include <c10/core/TensorOptions.h>
#include <c10/util/Deprecated.h>
#include <optional>
#include <string_view>



#include <ATen/ops/_cudnn_attention_forward_ops.h>

namespace at {


// aten::_cudnn_attention_forward(Tensor query, Tensor key, Tensor value, Tensor? attn_bias, Tensor? cum_seq_q, Tensor? cum_seq_k, SymInt max_q, SymInt max_k, bool compute_log_sumexp, float dropout_p=0.0, bool is_causal=False, bool return_debug_mask=False, *, float? scale=None) -> (Tensor output, Tensor logsumexp, Tensor cum_seq_q, Tensor cum_seq_k, SymInt max_q, SymInt max_k, Tensor philox_seed, Tensor philox_offset, Tensor debug_attn_mask)
inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt,at::Tensor,at::Tensor,at::Tensor> _cudnn_attention_forward(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_bias, const ::std::optional<at::Tensor> & cum_seq_q, const ::std::optional<at::Tensor> & cum_seq_k, int64_t max_q, int64_t max_k, bool compute_log_sumexp, double dropout_p=0.0, bool is_causal=false, bool return_debug_mask=false, ::std::optional<double> scale=::std::nullopt) {
    return at::_ops::_cudnn_attention_forward::call(query, key, value, attn_bias, cum_seq_q, cum_seq_k, max_q, max_k, compute_log_sumexp, dropout_p, is_causal, return_debug_mask, scale);
}
namespace symint {
  template <typename T, typename = std::enable_if_t<std::is_same_v<T, int64_t>>>
  ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt,at::Tensor,at::Tensor,at::Tensor> _cudnn_attention_forward(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_bias, const ::std::optional<at::Tensor> & cum_seq_q, const ::std::optional<at::Tensor> & cum_seq_k, int64_t max_q, int64_t max_k, bool compute_log_sumexp, double dropout_p=0.0, bool is_causal=false, bool return_debug_mask=false, ::std::optional<double> scale=::std::nullopt) {
    return at::_ops::_cudnn_attention_forward::call(query, key, value, attn_bias, cum_seq_q, cum_seq_k, max_q, max_k, compute_log_sumexp, dropout_p, is_causal, return_debug_mask, scale);
  }
}

// aten::_cudnn_attention_forward(Tensor query, Tensor key, Tensor value, Tensor? attn_bias, Tensor? cum_seq_q, Tensor? cum_seq_k, SymInt max_q, SymInt max_k, bool compute_log_sumexp, float dropout_p=0.0, bool is_causal=False, bool return_debug_mask=False, *, float? scale=None) -> (Tensor output, Tensor logsumexp, Tensor cum_seq_q, Tensor cum_seq_k, SymInt max_q, SymInt max_k, Tensor philox_seed, Tensor philox_offset, Tensor debug_attn_mask)
inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt,at::Tensor,at::Tensor,at::Tensor> _cudnn_attention_forward_symint(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_bias, const ::std::optional<at::Tensor> & cum_seq_q, const ::std::optional<at::Tensor> & cum_seq_k, c10::SymInt max_q, c10::SymInt max_k, bool compute_log_sumexp, double dropout_p=0.0, bool is_causal=false, bool return_debug_mask=false, ::std::optional<double> scale=::std::nullopt) {
    return at::_ops::_cudnn_attention_forward::call(query, key, value, attn_bias, cum_seq_q, cum_seq_k, max_q, max_k, compute_log_sumexp, dropout_p, is_causal, return_debug_mask, scale);
}
namespace symint {
  template <typename T, typename = std::enable_if_t<std::is_same_v<T, c10::SymInt>>>
  ::std::tuple<at::Tensor,at::Tensor,at::Tensor,at::Tensor,c10::SymInt,c10::SymInt,at::Tensor,at::Tensor,at::Tensor> _cudnn_attention_forward(const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const ::std::optional<at::Tensor> & attn_bias, const ::std::optional<at::Tensor> & cum_seq_q, const ::std::optional<at::Tensor> & cum_seq_k, c10::SymInt max_q, c10::SymInt max_k, bool compute_log_sumexp, double dropout_p=0.0, bool is_causal=false, bool return_debug_mask=false, ::std::optional<double> scale=::std::nullopt) {
    return at::_ops::_cudnn_attention_forward::call(query, key, value, attn_bias, cum_seq_q, cum_seq_k, max_q, max_k, compute_log_sumexp, dropout_p, is_causal, return_debug_mask, scale);
  }
}

}