未验证 提交 0daa5c97 编写于 作者: Z Zeng Jinle 提交者: GitHub

Make leaky relu inplacable (#19676)

* make leaky relu inplacable, test=develop

* force add unittests to pass coverage, test=develop
上级 c308c88d
develop 1.8.5 2.0.1-rocm-post 2.4.1 Ligoml-patch-1 OliverLPH-patch-1 OliverLPH-patch-2 PaddlePM-patch-1 PaddlePM-patch-2 ZHUI-patch-1 add_default_att add_kylinv10 add_model_benchmark_ci add_some_yaml_config addfile all_new_design_exec ascendrc ascendrelease bugfix-eval-frame-leakgae cherry-pick-fix-customOP-random-fail cherry_undefined_var compile_windows cp_2.4_fix_numpy delete_2.0.1-rocm-post delete_add_default_att delete_all_new_design_exec delete_ascendrc delete_compile_windows delete_delete_addfile delete_disable_iterable_dataset_unittest delete_fix_dataloader_memory_leak delete_fix_imperative_dygraph_error delete_fix_retry_ci delete_fix_undefined_var delete_improve_sccache delete_paddle_tiny_install delete_paralleltest delete_prv-disable-more-cache delete_revert-31068-fix_conv3d_windows delete_revert-31562-mean delete_revert-33630-bug-fix delete_revert-34159-add_npu_bce_logical_dev delete_revert-34910-spinlocks_for_allocator delete_revert-35069-revert-34910-spinlocks_for_allocator delete_revert-36057-dev/read_flags_in_ut dingjiaweiww-patch-1 disable_iterable_dataset_unittest dy2static enable_eager_model_test final_state_gen_python_c final_state_intermediate fix-numpy-issue fix-run-program-grad-node-mem fix_check fix_concat_slice fix_custom_device_copy_sync fix_dataloader_memory_leak fix_dlpack_for fix_imperative_dygraph_error fix_newexe_gc fix_npu_ci fix_op_flops fix_retry_ci fix_rnn_docs fix_tensor_type fix_undefined_var fix_var_stop_gradient_error fixiscan fixiscan1 fixiscan2 fixiscan3 github/fork/123malin/netifaces github/fork/123malin/tdm_abacus github/fork/AshburnLee/dev_unique github/fork/ForFishes/fix_memory_matmul github/fork/ForFishes/rm_fluid github/fork/LielinJiang/move-2.0-api github/fork/LielinJiang/visual-dl-cb github/fork/LiuChiachi/add-transformer-generate-square-subsequent-mask-api github/fork/LiuChiachi/fix-example-code-for-hapi-Model github/fork/LiuChiachi/remove-input-requirment-in-dygraph-Model github/fork/MrChengmo/fix_ps_profiler github/fork/MrChengmo/update_ps_heter github/fork/PWhiddy/patch-1 github/fork/Shixiaowei02/dev/save_load_upgrade github/fork/TCChenlong/fix_hapi github/fork/TCChenlong/fix_inden github/fork/Thunderbrook/xpu_slice github/fork/XieYunshen/disable_ut_test_parallel_executor_fetch_isolated_var github/fork/XieYunshen/disable_ut_test_parallel_executor_fetch_isolated_var_2 github/fork/XieYunshen/disable_ut_test_parallel_executor_fetch_isolated_var_3 github/fork/XieYunshen/timeout_20S_ut github/fork/ZeyuChen/remove-nltk github/fork/arlesniak/arlesniak/selective__mkldnn_flags github/fork/baiyfbupt/code_doc_mig github/fork/chalsliu/set_timeout github/fork/chen-zhiyu/develop github/fork/chenwhql/ci/try_to_find_test_buffer_shared_memory_reuse_pass_error github/fork/chenwhql/dygraph/remove_scale_loss_and_apply_collective_grads github/fork/chenwhql/saveload/add_get_inference_program github/fork/chenwhql/saveload/remove_save_load_config github/fork/cryoco/pass-compatibility-trt github/fork/danleifeng/isempty_api2.0 github/fork/frankwhzhang/api_transfer github/fork/hbwx24/error_msg/cuda_kernel_error_msg github/fork/heavengate/cherry_yolo_box github/fork/heavengate/update_yolo_box github/fork/iclementine/rnn_fix github/fork/iducn/testestse github/fork/jczaja/prv-25537-fix github/fork/jeff41404/release/1.8 github/fork/jiweibo/api_2.0 github/fork/jiweibo/fix_lite_resnet50_test github/fork/juncaipeng/fix_doc_1 github/fork/lfchener/sample_code github/fork/littletomatodonkey/fix_reg_doc github/fork/liym27/dy2stat_update_assign_to_rc20 github/fork/luotao1/profiler_ut github/fork/mapingshuo/add_wait github/fork/mapingshuo/doc_2.0 github/fork/mapingshuo/zero-0.5 github/fork/miraiwk/dev github/fork/pangyoki/add-Categorical-class-branch github/fork/pangyoki/add-multinomial-op-branch github/fork/pangyoki/fix-test_distritbution-CI github/fork/qjing666/doublegrad github/fork/qjing666/fix_hdfs_download github/fork/sandyhouse/add_gather_etc github/fork/sandyhouse/add_send_recv_alltoall_etc github/fork/sandyhouse/pipeline_exe_run github/fork/seiriosPlus/feature/large_scale_kv_save_delta github/fork/seiriosPlus/fix/paddle_errors_fix github/fork/seiriosPlus/fix/paddle_op_errors github/fork/shangzhizhou/fix_test_activation_op_random_bug github/fork/smallv0221/yxp0924 github/fork/smallv0221/yxp0925 github/fork/swtkiwi/del-matplotlib github/fork/tianshuo78520a/kunlun_test github/fork/tianshuo78520a/update_dockerfile github/fork/wanghaoshuang/bert_fuse github/fork/wanghaoshuang/label_smooth github/fork/wanghuancoder/develop_CUDASynchronize github/fork/wanghuancoder/develop_Layer_doc github/fork/wanghuancoder/develop_ParameterList_doc github/fork/wanghuancoder/develop_Sequential_doc github/fork/wanghuancoder/develop_bilinear_tensor_product github/fork/wanghuancoder/develop_coverage_build_sh github/fork/wanghuancoder/develop_in_dynamic_mode_doc github/fork/wanghuancoder/develop_unique_name_doc github/fork/wangxicoding/fleet_meta_combine github/fork/wawltor/error_message_fix_5 github/fork/willthefrog/remove_l2_norm github/fork/windstamp/momentum_op github/fork/windstamp/mv_op_5 github/fork/windstamp/normal_api github/fork/wojtuss/wojtuss/fusion_gru_quantization github/fork/wojtuss/wojtuss/quantization-with-shift github/fork/wzzju/fix_err_info github/fork/wzzju/pure_fp16 github/fork/xiemoyuan/op_error_message github/fork/xiemoyuan/optimize_error_message github/fork/yaoxuefeng6/fix_doc github/fork/yaoxuefeng6/mod_dataset_v2 github/fork/yongqiangma/lod github/fork/ysh329/fix-clip-by-norm-error github/fork/ysh329/fix-error-clip-by-value github/fork/yukavio/error_info github/fork/zhangting2020/conv_filter_grad github/fork/zhangting2020/is_compile_with_cuda github/fork/zhangting2020/place_doc github/fork/zhangting2020/program github/fork/zhhsplendid/fix_any github/fork/zhhsplendid/refine_api2 github/fork/zhhsplendid/refine_api2_test github/fork/zhhsplendid/refine_api_test_ptb_lm github/fork/zhhsplendid/refine_api_test_resnet github/fork/zhhsplendid/refine_api_test_simnet github/fork/zhiqiu/dev/refine_initializer github/fork/zhiqiu/dev/remove_inplace_argument github/fork/zlsh80826/nvinfer_plugin_var_len_cuda11 hack_event improve_sccache incuabte/new_frl incubate/frl_train_eval incubate/infrt incubate/new_frl incubate/new_frl_rc incubate/stride inplace_addto layer_norm make_flag_adding_easier matmul_double_grad move_embedding_to_phi move_histogram_to_pten move_sgd_to_phi move_slice_to_pten move_temporal_shift_to_phi move_yolo_box_to_phi npu_fix_alloc numel operator_opt paddle_tiny_install paralleltest pass-compile-eval-frame preln_ernie prv-disable-more-cache prv-md-even-more prv-onednn-2.5 prv-reshape-mkldnn-ut2 pten_tensor_refactor release-deleted/2.5 release-rc/2.5 release/1.6 release/1.7 release/1.8 release/2.0 release/2.0-alpha release/2.0-beta release/2.0-rc release/2.0-rc1 release/2.1 release/2.2 release/2.3 release/2.3-fc-ernie-fix release/2.4 release/2.5 release/llm_2.5 revert-24981-add_device_attr_for_regulization revert-26856-strategy_example2 revert-27520-disable_pr revert-31068-fix_conv3d_windows revert-31562-mean revert-32290-develop-hardlabel revert-33037-forci revert-33475-fix_cifar_label_dimension revert-33630-bug-fix revert-34159-add_npu_bce_logical_dev revert-34406-add_copy_from_tensor revert-34910-spinlocks_for_allocator revert-35069-revert-34910-spinlocks_for_allocator revert-36057-dev/read_flags_in_ut revert-36201-refine_fast_threaded_ssa_graph_executor revert-36985-add_license revert-37318-refactor_dygraph_to_eager revert-37926-eager_coreops_500 revert-37956-revert-37727-pylayer_support_tuple revert-38100-mingdong revert-38301-allocation_rearrange_pr revert-38703-numpy_bf16_package_reupload revert-38732-remove_useless_header_in_elementwise_mul_grad revert-38959-Reduce_Grad revert-39143-adjust_empty revert-39227-move_trace_op_to_pten revert-39268-dev/remove_concat_fluid_kernel revert-40170-support_partial_grad revert-41056-revert-40727-move_some_activaion_to_phi revert-41065-revert-40993-mv_ele_floordiv_pow revert-41068-revert-40790-phi_new revert-41944-smaller_inference_api_test revert-42149-do-not-reset-default-stream-for-stream-safe-cuda-allocator revert-43155-fix_ut_tempfile revert-43882-revert-41944-smaller_inference_api_test revert-45808-phi/simplify_size_op revert-46827-deform_comment revert-47325-remove_cudnn_hardcode revert-47645-add_npu_storage_dims revert-48815-set_free_when_no_cache_hit_default_value_true revert-49499-test_ninja_on_ci revert-49654-prim_api_gen revert-49673-modify_get_single_cov revert-49763-fix_static_composite_gen revert-50158-fix_found_inf_bug_for_custom_optimizer revert-50188-refine_optimizer_create_accumulators revert-50335-fix_optminizer_set_auxiliary_var_bug revert-51676-flag_delete revert-51850-fix_softmaxce_dev revert-52175-dev_peak_memory revert-52186-deve revert-52523-test_py38 revert-52912-develop revert-53248-set_cmake_policy revert-54029-fix_windows_compile_bug revert-54068-support_translating_op_attribute revert-54214-modify_cmake_dependencies revert-54370-offline_pslib revert-54391-fix_cmake_md5error revert-54411-fix_cpp17_compile revert-54466-offline_pslib revert-54480-cmake-rocksdb revert-55568-fix_BF16_bug1 revert-56328-new_ir_support_vector_type_place_transfer revert-56366-fix_openssl_bug revert-56545-revert-56366-fix_openssl_bug revert-56620-fix_new_ir_ocr_bug revert-56925-check_inputs_grad_semantic revert-57005-refine_stride_flag rocm_dev_0217 sd_conv_linear_autocast semi-auto/rule-base support-0D-sort support_weight_transpose test_benchmark_ci test_feature_precision_test_c test_for_Filtetfiles test_model_benchmark test_model_benchmark_ci zhiqiu-patch-1 v2.5.1 v2.5.0 v2.5.0-rc1 v2.5.0-rc0 v2.4.2 v2.4.1 v2.4.0 v2.4.0-rc0 v2.3.2 v2.3.1 v2.3.0 v2.3.0-rc0 v2.2.2 v2.2.1 v2.2.0 v2.2.0-rc0 v2.2.0-bak0 v2.1.3 v2.1.2 v2.1.1 v2.1.0 v2.1.0-rc0 v2.0.2 v2.0.1 v2.0.0 v2.0.0-rc1 v2.0.0-rc0 v2.0.0-beta0 v2.0.0-alpha0 v1.8.5 v1.8.4 v1.8.3 v1.8.2 v1.8.1 v1.8.0 v1.7.2 v1.7.1 v1.7.0 v1.6.3 v1.6.2 v1.6.1 v1.6.0 v1.6.0-rc0
无相关合并请求
......@@ -250,6 +250,8 @@ ir::Graph *ParallelExecutorPrivate::ApplyMemoryOptimizePass(ir::Graph *graph) {
VLOG(10) << "Start to apply buffer_shared_inplace_pass";
graph = inplace_pass->Apply(graph);
VLOG(10) << "buffer_shared_inplace_pass Applied";
LOG(INFO) << "Inplace strategy is enabled, when "
"build_strategy.enable_inplace = True";
}
/**
......@@ -278,6 +280,9 @@ ir::Graph *ParallelExecutorPrivate::ApplyMemoryOptimizePass(ir::Graph *graph) {
VLOG(10) << "Start to apply buffer_shared_cross_op_memory_reuse_pass";
graph = cross_op_memory_reuse_pass->Apply(graph);
VLOG(10) << "buffer_shared_cross_op_memory_reuse_pass Applied";
LOG(INFO) << "Cross op memory reuse strategy is enabled, when "
"build_strategy.memory_optimize = True or garbage collection "
"strategy is disabled, which is not recommended";
}
if (!is_gc_enabled) {
......
......@@ -116,6 +116,11 @@ cc_test(strided_memcpy_test SRCS strided_memcpy_test.cc DEPS tensor memory)
cc_test(save_load_op_test SRCS save_load_op_test.cc DEPS save_op load_op)
cc_test(save_load_combine_op_test SRCS save_load_combine_op_test.cc DEPS save_combine_op load_combine_op)
nv_test(dropout_op_test SRCS dropout_op_test.cc DEPS dropout_op tensor)
if (WITH_GPU)
nv_test(test_leaky_relu_grad_grad_functor SRCS test_leaky_relu_grad_grad_functor.cc test_leaky_relu_grad_grad_functor.cu DEPS tensor device_context eigen3)
else()
cc_test(test_leaky_relu_grad_grad_functor SRCS test_leaky_relu_grad_grad_functor.cc DEPS tensor device_context eigen3)
endif()
if (WITH_PYTHON)
cc_library(py_func_op SRCS py_func_op.cc DEPS op_registry python pybind)
......
......@@ -716,8 +716,8 @@ class LeakyReluDoubleGradMaker
std::unique_ptr<::paddle::framework::OpDesc> Apply() const override {
auto* op = new ::paddle::framework::OpDesc();
op->SetType("leaky_relu_grad_grad");
// input1: X
op->SetInput("X", Input("X"));
// input1: Out
op->SetInput("Out", Input("Out"));
// X@GRAD@GRAD: ddx
op->SetInput("DDX", OutputGrad(framework::GradVarName("X")));
op->SetAttrMap(Attrs());
......
......@@ -463,8 +463,8 @@ struct HardShrinkFunctor : public BaseActivationFunctor<T> {
}
template <typename Device, typename X, typename Out>
void operator()(Device d, X x, Out out) const {
auto temp1 = (x < static_cast<T>(threshold * -1)).template cast<T>().eval();
auto temp2 = (x > static_cast<T>(threshold)).template cast<T>().eval();
auto temp1 = (x < static_cast<T>(threshold * -1)).template cast<T>();
auto temp2 = (x > static_cast<T>(threshold)).template cast<T>();
out.device(d) = x * (temp1 + temp2);
}
};
......@@ -480,8 +480,8 @@ struct HardShrinkGradFunctor : public BaseActivationFunctor<T> {
template <typename Device, typename X, typename Out, typename dOut,
typename dX>
void operator()(Device d, X x, Out out, dOut dout, dX dx) const {
auto temp1 = (x < static_cast<T>(threshold * -1)).template cast<T>().eval();
auto temp2 = (x > static_cast<T>(threshold)).template cast<T>().eval();
auto temp1 = (x < static_cast<T>(threshold * -1)).template cast<T>();
auto temp2 = (x > static_cast<T>(threshold)).template cast<T>();
dx.device(d) = dout * (temp1 + temp2).template cast<T>();
}
......@@ -500,8 +500,8 @@ struct SoftShrinkFunctor : public BaseActivationFunctor<T> {
template <typename Device, typename X, typename Out>
void operator()(Device d, X x, Out out) const {
auto lambdaT = static_cast<T>(lambda);
auto temp1 = (x > lambdaT).template cast<T>().eval();
auto temp2 = (x < -lambdaT).template cast<T>().eval();
auto temp1 = (x > lambdaT).template cast<T>();
auto temp2 = (x < -lambdaT).template cast<T>();
out.device(d) = temp1 * (x - lambdaT) + temp2 * (x + lambdaT);
}
};
......@@ -516,8 +516,8 @@ struct SoftShrinkGradFunctor : public BaseActivationFunctor<T> {
typename dX>
void operator()(Device d, X x, Out out, dOut dout, dX dx) const {
auto lambdaT = static_cast<T>(lambda);
auto temp1 = (x > lambdaT).template cast<T>().eval();
auto temp2 = (x < -lambdaT).template cast<T>().eval();
auto temp1 = (x > lambdaT).template cast<T>();
auto temp2 = (x < -lambdaT).template cast<T>();
dx.device(d) = dout * (temp1 + temp2).template cast<T>();
}
......@@ -1043,7 +1043,7 @@ struct SoftReluGradFunctor : public BaseActivationFunctor<T> {
typename dX>
void operator()(Device d, X x, Out out, dOut dout, dX dx) const {
auto tmp = static_cast<T>(threshold);
auto temp = ((out > -tmp) * (out < tmp)).template cast<T>().eval();
auto temp = ((out > -tmp) * (out < tmp)).template cast<T>();
dx.device(d) = dout * (static_cast<T>(1) - (-out).exp()) * temp;
}
......@@ -1072,13 +1072,13 @@ struct LeakyReluGradFunctor : public BaseActivationFunctor<T> {
template <typename Device, typename X, typename Out, typename dOut,
typename dX>
void operator()(Device d, X x, Out out, dOut dout, dX dx) const {
auto temp1 = static_cast<T>(alpha) *
(x < static_cast<T>(0)).template cast<T>().eval();
auto temp2 = (x >= static_cast<T>(0)).template cast<T>().eval();
auto temp1 =
static_cast<T>(alpha) * (out < static_cast<T>(0)).template cast<T>();
auto temp2 = (out >= static_cast<T>(0)).template cast<T>();
dx.device(d) = dout * (temp1 + temp2).template cast<T>();
}
static constexpr ActBwdOpFwdDeps FwdDeps() { return kDepX; }
static constexpr ActBwdOpFwdDeps FwdDeps() { return kDepOut; }
};
template <typename T>
......@@ -1413,19 +1413,19 @@ struct LeakyReluGradGradFunctor : public BaseActivationFunctor<T> {
const framework::Tensor* Out, const framework::Tensor* ddX,
framework::Tensor* ddOut, framework::Tensor* dOut,
framework::Tensor* dX) const {
auto* d = dev.eigen_device();
auto ddx = framework::EigenVector<T>::Flatten(detail::Ref(ddX));
auto x = framework::EigenVector<T>::Flatten(detail::Ref(X));
if (ddOut) {
auto* d = dev.eigen_device();
auto ddx = framework::EigenVector<T>::Flatten(detail::Ref(ddX));
auto out = framework::EigenVector<T>::Flatten(detail::Ref(Out));
auto ddout = framework::EigenVector<T>::Flatten(detail::Ref(ddOut));
ddout.device(*d) = ddx *
((x >= static_cast<T>(0)).template cast<T>().eval() +
static_cast<T>(alpha) *
(x < static_cast<T>(0)).template cast<T>().eval())
.template cast<T>();
ddout.device(*d) =
ddx *
((out >= static_cast<T>(0)).template cast<T>() +
static_cast<T>(alpha) * (out < static_cast<T>(0)).template cast<T>())
.template cast<T>();
}
}
static constexpr ActBwdOpFwdDeps FwdDeps() { return kDepX; }
static constexpr ActBwdOpFwdDeps FwdDeps() { return kDepOut; }
};
template <typename T>
......
// Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
#include "paddle/fluid/operators/test_leaky_relu_grad_grad_functor.h"
namespace paddle {
namespace operators {
TEST(leaky_relu_grad_grad, test_cpu) {
ASSERT_TRUE(
TestLeakyReluGradGradMain<float>({32, 64}, platform::CPUPlace(), 0.02));
}
} // namespace operators
} // namespace paddle
// Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
#include "paddle/fluid/operators/test_leaky_relu_grad_grad_functor.h"
namespace paddle {
namespace operators {
TEST(leaky_relu_grad_grad, test_gpu) {
ASSERT_TRUE(
TestLeakyReluGradGradMain<float>({32, 64}, platform::CUDAPlace(0), 0.15));
}
} // namespace operators
} // namespace paddle
// Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
#pragma once
#include <algorithm>
#include <random>
#include "gtest/gtest.h"
#include "paddle/fluid/operators/activation_op.h"
#include "paddle/fluid/platform/for_range.h"
namespace paddle {
namespace operators {
template <typename T>
static void InitRandom(framework::Tensor *tensor,
const platform::Place &place) {
framework::Tensor cpu_tensor;
auto *cpu_ptr =
cpu_tensor.mutable_data<T>(tensor->dims(), platform::CPUPlace());
int64_t numel = cpu_tensor.numel();
std::mt19937 engine;
std::uniform_real_distribution<T> dist(static_cast<T>(-2.0),
static_cast<T>(2.0));
for (int64_t i = 0; i < numel; ++i) {
cpu_ptr[i] = dist(engine);
}
framework::TensorCopySync(cpu_tensor, place, tensor);
}
template <typename T>
struct LeakyReluGradGradEachElementFunctor {
LeakyReluGradGradEachElementFunctor(const T *ddx, const T *out, T alpha,
T *ddout)
: ddx_(ddx), out_(out), alpha_(alpha), ddout_(ddout) {}
HOSTDEVICE void operator()(int idx) {
if (out_[idx] >= 0) {
ddout_[idx] = ddx_[idx];
} else {
ddout_[idx] = ddx_[idx] * alpha_;
}
}
const T *ddx_;
const T *out_;
T alpha_;
T *ddout_;
};
template <typename T>
static bool TestLeakyReluGradGradMain(const framework::DDim &dim,
const platform::Place &place,
float alpha) {
LeakyReluGradGradFunctor<T> functor;
functor.alpha = alpha;
auto &dev_ctx = *platform::DeviceContextPool::Instance().Get(place);
framework::Tensor *x = nullptr;
framework::Tensor *dout = nullptr;
framework::Tensor *dx = nullptr;
framework::Tensor out;
out.Resize(dim);
InitRandom<T>(&out, place);
framework::Tensor ddx;
ddx.Resize(dim);
InitRandom<T>(&ddx, place);
framework::Tensor ddout;
ddout.Resize(dim);
InitRandom<T>(&ddout, place);
framework::Tensor ddout_actual;
ddout_actual.mutable_data<T>(dim, place);
LeakyReluGradGradEachElementFunctor<T> actual_functor(
ddx.data<T>(), out.data<T>(), static_cast<T>(alpha),
ddout_actual.data<T>());
int64_t limit = out.numel();
#ifdef __NVCC__
if (platform::is_gpu_place(place)) {
auto &cuda_dev_ctx = dynamic_cast<platform::CUDADeviceContext &>(dev_ctx);
functor(cuda_dev_ctx, x, &out, &ddx, &ddout, dout, dx);
platform::ForRange<platform::CUDADeviceContext> for_range(cuda_dev_ctx,
limit);
for_range(actual_functor);
} else {
#endif
auto &cpu_dev_ctx = dynamic_cast<platform::CPUDeviceContext &>(dev_ctx);
functor(cpu_dev_ctx, x, &out, &ddx, &ddout, dout, dx);
platform::ForRange<platform::CPUDeviceContext> for_range(cpu_dev_ctx,
limit);
for_range(actual_functor);
#ifdef __NVCC__
}
#endif
dev_ctx.Wait();
framework::Tensor ddout_cpu, ddout_actual_cpu;
framework::TensorCopySync(ddout, platform::CPUPlace(), &ddout_cpu);
framework::TensorCopySync(ddout_actual, platform::CPUPlace(),
&ddout_actual_cpu);
bool is_equal = std::equal(ddout_cpu.data<T>(), ddout_cpu.data<T>() + limit,
ddout_actual_cpu.data<T>());
return is_equal;
}
} // namespace operators
} // namespace paddle
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册
反馈
建议
客服 返回
顶部