1#pragma once
2
3// @generated by torchgen/gen.py from Function.h
4
5#include <ATen/Context.h>
6#include <ATen/DeviceGuard.h>
7#include <ATen/TensorUtils.h>
8#include <ATen/TracerMode.h>
9#include <ATen/core/Generator.h>
10#include <ATen/core/Reduction.h>
11#include <ATen/core/Tensor.h>
12#include <c10/core/Scalar.h>
13#include <c10/core/Storage.h>
14#include <c10/core/TensorOptions.h>
15#include <c10/util/Deprecated.h>
16#include <c10/util/Optional.h>
17
18
19
20#include <ATen/ops/_scaled_dot_product_flash_attention_backward_ops.h>
21
22namespace at {
23
24
25// aten::_scaled_dot_product_flash_attention_backward(Tensor grad_out, Tensor query, Tensor key, Tensor value, Tensor out, Tensor logsumexp, Tensor cum_seq_q, Tensor cum_seq_k, int max_q, int max_k, float dropout_p, bool is_causal, int philox_seed, int philox_offset) -> (Tensor grad_query, Tensor grad_key, Tensor grad_value)
26inline ::std::tuple<at::Tensor,at::Tensor,at::Tensor> _scaled_dot_product_flash_attention_backward(const at::Tensor & grad_out, const at::Tensor & query, const at::Tensor & key, const at::Tensor & value, const at::Tensor & out, const at::Tensor & logsumexp, const at::Tensor & cum_seq_q, const at::Tensor & cum_seq_k, int64_t max_q, int64_t max_k, double dropout_p, bool is_causal, int64_t philox_seed, int64_t philox_offset) {
27 return at::_ops::_scaled_dot_product_flash_attention_backward::call(grad_out, query, key, value, out, logsumexp, cum_seq_q, cum_seq_k, max_q, max_k, dropout_p, is_causal, philox_seed, philox_offset);
28}
29
30}
31