data_norm_op.cc 32.2 KB
Newer Older
H
heqiaozhi 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/* Copyright (c) 2016 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/data_norm_op.h"
P
phlrain 已提交
16
#include <memory>
H
heqiaozhi 已提交
17 18
#include <string>
#include "paddle/fluid/framework/data_layout.h"
H
heqiaozhi 已提交
19 20 21
#ifdef PADDLE_WITH_MKLDNN
#include "paddle/fluid/platform/mkldnn_helper.h"
#endif
22
#include "paddle/fluid/framework/op_version_registry.h"
H
heqiaozhi 已提交
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47

namespace paddle {
namespace operators {

using Tensor = framework::Tensor;
using LoDTensor = framework::LoDTensor;
using DataLayout = framework::DataLayout;

template <typename T>
using EigenArrayMap =
    Eigen::Map<Eigen::Array<T, Eigen::Dynamic, Eigen::Dynamic>>;
template <typename T>
using ConstEigenArrayMap =
    Eigen::Map<const Eigen::Array<T, Eigen::Dynamic, Eigen::Dynamic>>;
template <typename T>
using EigenVectorArrayMap = Eigen::Map<Eigen::Array<T, Eigen::Dynamic, 1>>;
template <typename T>
using ConstEigenVectorArrayMap =
    Eigen::Map<const Eigen::Array<T, Eigen::Dynamic, 1>>;

class DataNormOp : public framework::OperatorWithKernel {
 public:
  using framework::OperatorWithKernel::OperatorWithKernel;

  void InferShape(framework::InferShapeContext *ctx) const override {
48 49 50 51 52 53 54 55 56
    OP_INOUT_CHECK(ctx->HasInput("X"), "Input", "X", "DataNorm");
    OP_INOUT_CHECK(ctx->HasInput("BatchSize"), "Input", "BatchSize",
                   "DataNorm");
    OP_INOUT_CHECK(ctx->HasInput("BatchSum"), "Input", "BatchSum", "DataNorm");
    OP_INOUT_CHECK(ctx->HasInput("BatchSquareSum"), "Input", "BatchSquareSum",
                   "DataNorm");
    OP_INOUT_CHECK(ctx->HasOutput("Means"), "Output", "Means", "DataNorm");
    OP_INOUT_CHECK(ctx->HasOutput("Scales"), "Output", "Scales", "DataNorm");
    OP_INOUT_CHECK(ctx->HasOutput("Y"), "Output", "Y", "DataNorm");
57 58 59 60 61 62 63 64 65 66 67
    bool enable_scale_and_shift =
        ctx->Attrs().Get<bool>("enable_scale_and_shift");
    if (enable_scale_and_shift) {
      PADDLE_ENFORCE_EQ(
          ctx->HasInput("scale_w"), true,
          platform::errors::InvalidArgument(
              "Input(scale_w) of DataNormOp should not be null."));
      PADDLE_ENFORCE_EQ(ctx->HasInput("bias"), true,
                        platform::errors::InvalidArgument(
                            "Input(bias) of DataNormOp should not be null."));
    }
H
heqiaozhi 已提交
68 69 70 71 72

    const auto x_dims = ctx->GetInputDim("X");
    const DataLayout data_layout = framework::StringToDataLayout(
        ctx->Attrs().Get<std::string>("data_layout"));

73 74 75
    PADDLE_ENFORCE_EQ(x_dims.size() >= 2 && x_dims.size() <= 5, true,
                      platform::errors::InvalidArgument(
                          "Input X must have 2 to 5 dimensions."));
H
heqiaozhi 已提交
76 77 78 79 80

    const int64_t C =
        (data_layout == DataLayout::kNCHW ? x_dims[1]
                                          : x_dims[x_dims.size() - 1]);

81 82 83 84 85 86 87 88 89
    PADDLE_ENFORCE_EQ(ctx->GetInputDim("BatchSize").size(), 1UL,
                      platform::errors::InvalidArgument(
                          "The input dim of BatchSize shouold be 1"));
    PADDLE_ENFORCE_EQ(ctx->GetInputDim("BatchSum").size(), 1UL,
                      platform::errors::InvalidArgument(
                          "The input dim of BatchSum shouold be 1"));
    PADDLE_ENFORCE_EQ(ctx->GetInputDim("BatchSquareSum").size(), 1UL,
                      platform::errors::InvalidArgument(
                          "The input dim of BatchSquareSum shouold be 1"));
P
phlrain 已提交
90
    if (ctx->IsRuntime()) {
91 92 93 94 95 96 97 98 99
      PADDLE_ENFORCE_EQ(ctx->GetInputDim("BatchSize")[0], C,
                        platform::errors::InvalidArgument(
                            "The input dim[0] of BatchSize shouold be C"));
      PADDLE_ENFORCE_EQ(ctx->GetInputDim("BatchSum")[0], C,
                        platform::errors::InvalidArgument(
                            "The input dim[0] of BatchSum shouold be C"));
      PADDLE_ENFORCE_EQ(ctx->GetInputDim("BatchSquareSum")[0], C,
                        platform::errors::InvalidArgument(
                            "The input dim[0] of BatchSqureSum shouold be C"));
P
phlrain 已提交
100
    }
H
heqiaozhi 已提交
101

102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140
    if (enable_scale_and_shift) {
      auto scale_dim = ctx->GetInputDim("scale_w");
      auto bias_dim = ctx->GetInputDim("bias");

      PADDLE_ENFORCE_EQ(
          scale_dim.size(), 1UL,
          platform::errors::InvalidArgument("the dimensionof scale"
                                            "must equal to 1. But received: "
                                            "the shape of scale is [%s], "
                                            "the dimensionof scale is [%d]",
                                            scale_dim, scale_dim.size()));
      PADDLE_ENFORCE_EQ(
          bias_dim.size(), 1UL,
          platform::errors::InvalidArgument("the dimension of bias"
                                            "must equal to 1. But received: "
                                            "the shape of bias is [%s],"
                                            "the dimension of bias is [%d]",
                                            bias_dim, bias_dim.size()));

      bool check = true;
      if ((!ctx->IsRuntime()) && (framework::product(scale_dim) <= 0 ||
                                  framework::product(bias_dim) <= 0)) {
        check = false;
      }

      if (check) {
        PADDLE_ENFORCE_EQ(scale_dim[0], C,
                          platform::errors::InvalidArgument(
                              "the shape of scale must equal to [%d]"
                              "But received: the shape of scale is [%d]",
                              C, scale_dim[0]));
        PADDLE_ENFORCE_EQ(bias_dim[0], C,
                          platform::errors::InvalidArgument(
                              "the shape of bias must equal to [%d]"
                              "But received: the shape of bias is [%d]",
                              C, bias_dim[0]));
      }
    }

H
heqiaozhi 已提交
141 142 143 144 145 146 147 148 149
    ctx->SetOutputDim("Y", x_dims);
    ctx->SetOutputDim("Means", {C});
    ctx->SetOutputDim("Scales", {C});
    ctx->ShareLoD("X", "Y");
  }

 protected:
  framework::OpKernelType GetExpectedKernelType(
      const framework::ExecutionContext &ctx) const override {
150
    auto input_data_type = OperatorWithKernel::IndicateVarDataType(ctx, "X");
H
heqiaozhi 已提交
151 152 153 154 155 156 157
    // By default, the type of the scale, bias, mean,
    // and var tensors should both be float. (For float or float16 input tensor)
    // or double (For double input tensor).
    auto dn_param_type = framework::proto::VarType::FP32;
    if (input_data_type == framework::proto::VarType::FP64) {
      dn_param_type = framework::proto::VarType::FP64;
    }
158 159
    PADDLE_ENFORCE_EQ(dn_param_type,
                      OperatorWithKernel::IndicateVarDataType(ctx, "BatchSize"),
160 161
                      platform::errors::InvalidArgument(
                          "BatchSize input should be of float type"));
H
heqiaozhi 已提交
162
    PADDLE_ENFORCE_EQ(dn_param_type,
163
                      OperatorWithKernel::IndicateVarDataType(ctx, "BatchSum"),
164 165
                      platform::errors::InvalidArgument(
                          "BatchSum input should be of float type"));
166 167
    PADDLE_ENFORCE_EQ(dn_param_type, OperatorWithKernel::IndicateVarDataType(
                                         ctx, "BatchSquareSum"),
168 169
                      platform::errors::InvalidArgument(
                          "BatchSquareSum input should be of float type"));
H
heqiaozhi 已提交
170

171 172 173 174 175 176 177 178 179 180 181
    bool enable_scale_and_shift = ctx.Attr<bool>("enable_scale_and_shift");
    if (enable_scale_and_shift) {
      PADDLE_ENFORCE_EQ(dn_param_type,
                        OperatorWithKernel::IndicateVarDataType(ctx, "scale_w"),
                        platform::errors::InvalidArgument(
                            "scale_w input should be of float type"));
      PADDLE_ENFORCE_EQ(dn_param_type,
                        OperatorWithKernel::IndicateVarDataType(ctx, "bias"),
                        platform::errors::InvalidArgument(
                            "bias input should be of float type"));
    }
H
heqiaozhi 已提交
182 183 184
    // TODO(pzelazko-intel): enable MKLDNN layout when it's ready
    framework::LibraryType library = framework::LibraryType::kPlain;
    framework::DataLayout layout = framework::DataLayout::kAnyLayout;
H
heqiaozhi 已提交
185 186
#ifdef PADDLE_WITH_MKLDNN
    if (library == framework::LibraryType::kPlain &&
187
        this->CanMKLDNNBeUsed(ctx)) {
H
heqiaozhi 已提交
188 189 190 191
      library = framework::LibraryType::kMKLDNN;
      layout = framework::DataLayout::kMKLDNN;
    }
#endif
H
heqiaozhi 已提交
192 193 194 195 196 197 198 199 200 201 202 203 204

    return framework::OpKernelType(input_data_type, ctx.GetPlace(), layout,
                                   library);
  }
};

class DataNormOpMaker : public framework::OpProtoAndCheckerMaker {
 public:
  void Make() override {
    // AddAttr<bool>("is_test", "").SetDefault(false);
    AddAttr<float>("epsilon", "")
        .SetDefault(1e-4)
        .AddCustomChecker([](const float &epsilon) {
205 206 207
          PADDLE_ENFORCE_EQ(epsilon >= 0.0f && epsilon <= 0.001f, true,
                            platform::errors::InvalidArgument(
                                "'epsilon' should be between 0.0 and 0.001."));
H
heqiaozhi 已提交
208
        });
209 210 211 212
    AddAttr<int>("slot_dim",
                 "(int, default -1) Dimension of one slot if set, "
                 "when the input is concated by slot-wise embeddings")
        .SetDefault(-1);
H
hutuxian 已提交
213 214 215 216
    AddAttr<float>(
        "summary_decay_rate",
        "(float, default 0.9999999) The decay rate when update the summary")
        .SetDefault(0.9999999);
217 218 219 220 221 222 223 224 225 226 227 228 229
    AddAttr<bool>(
        "enable_scale_and_shift",
        "(bool, default false) Set to true to enable scale and shift such as "
        "batch_norm op")
        .SetDefault(false);
    AddInput("scale_w",
             "scale_w is a 1-dimensional tensor of size C "
             "that is applied to the output")
        .AsDispensable();
    AddInput("bias",
             "bias is a 1-dimensional tensor of size C "
             "that is applied to the output")
        .AsDispensable();
H
heqiaozhi 已提交
230
    AddAttr<std::string>("data_layout", "").SetDefault("NCHW");
H
hutuxian 已提交
231 232
    AddAttr<bool>("sync_stats", "(bool, default false) only used in multi-GPU")
        .SetDefault(false);
233 234 235
    AddAttr<bool>("use_mkldnn",
                  "(bool, default false) Only used in mkldnn kernel")
        .SetDefault(false);
H
heqiaozhi 已提交
236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278
    AddInput("X", "The input tensor");
    AddInput("BatchSize",
             "BatchSize is a 1-dimensional tensor of size C "
             "that is applied to the output");
    AddInput("BatchSum",
             "BatchSum is a 1-dimensional tensor of size C "
             "that is applied to the output");
    AddInput("BatchSquareSum",
             "The global BatchSquareSum (for training) or "
             "estimated BatchSquareSum (for testing)");
    AddOutput("Y", "result after normalization");
    AddOutput("Means",
              "Mean of the history data batch, "
              "will apply to output when training")
        .AsIntermediate();
    AddOutput("Scales",
              "Scales of the history data batch, "
              "will apply to output when training")
        .AsIntermediate();
    AddComment(R"DOC(
Data Normalization.

Can be used as a normalizer function for data
The required data format for this layer is one of the following:
1. NHWC `[batch, in_height, in_width, in_channels]`
2. NCHW `[batch, in_channels, in_height, in_width]`

)DOC");
  }
};

template <typename T>
class DataNormKernel<platform::CPUDeviceContext, T>
    : public framework::OpKernel<T> {
 public:
  void Compute(const framework::ExecutionContext &ctx) const override {
    // const bool is_test = ctx.Attr<bool>("is_test");
    const std::string data_layout_str = ctx.Attr<std::string>("data_layout");
    const DataLayout data_layout =
        framework::StringToDataLayout(data_layout_str);

    const auto *x = ctx.Input<Tensor>("X");
    const auto &x_dims = x->dims();
279 280
    PADDLE_ENFORCE_EQ(x_dims.size(), 2, platform::errors::InvalidArgument(
                                            "The Input dim size should be 2"));
H
heqiaozhi 已提交
281 282 283 284 285 286 287 288 289
    const int N = x_dims[0];
    const int C =
        (data_layout == DataLayout::kNCHW ? x_dims[1]
                                          : x_dims[x_dims.size() - 1]);
    auto *y = ctx.Output<Tensor>("Y");
    auto *mean_out = ctx.Output<Tensor>("Means");
    auto *scales = ctx.Output<Tensor>("Scales");

    // alloc memory
290
    T *y_data = y->mutable_data<T>(ctx.GetPlace());
H
heqiaozhi 已提交
291 292 293 294 295 296 297 298 299 300 301 302 303 304

    ConstEigenVectorArrayMap<T> b_size_arr(
        ctx.Input<Tensor>("BatchSize")->data<T>(), C);
    ConstEigenVectorArrayMap<T> b_sum_arr(
        ctx.Input<Tensor>("BatchSum")->data<T>(), C);
    ConstEigenVectorArrayMap<T> b_square_sum_arr(
        ctx.Input<Tensor>("BatchSquareSum")->data<T>(), C);
    EigenVectorArrayMap<T> means_arr(mean_out->mutable_data<T>(ctx.GetPlace()),
                                     C);
    EigenVectorArrayMap<T> scales_arr(scales->mutable_data<T>(ctx.GetPlace()),
                                      C);
    means_arr = b_sum_arr / b_size_arr;
    scales_arr = (b_size_arr / b_square_sum_arr).sqrt();

305 306
    const T *means_data = mean_out->data<T>();
    const T *x_data = x->data<T>();
307

308 309 310
    const T *scales_data = scales->data<T>();
    const int slot_dim = ctx.Attr<int>("slot_dim");
    T min_precision = 1e-7f;
H
heqiaozhi 已提交
311
    switch (data_layout) {
312
      case DataLayout::kNCHW:  // It's two dimensions, so make no difference
H
heqiaozhi 已提交
313
      case DataLayout::kNHWC: {
314 315
        // if slot_dim is set and batch size is larger than zero, we choose
        // to check if show number is zero, if so, skip normalization.
316 317
        if (slot_dim > 0 && N > 0 &&
            (!ctx.Attr<bool>("enable_scale_and_shift"))) {
318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337
          const int item_size = x->numel() / N;
          // location of show number in one embedding
          int offset = 0;
          for (int k = 0; k < N; ++k) {
            for (int i = 0; i < item_size; i += slot_dim) {
              if (x_data[offset + i] > -min_precision &&
                  x_data[offset + i] < min_precision) {
                // show = 0
                memset(y_data + offset + i, 0, sizeof(T) * slot_dim);
              } else {
                for (int j = i; j < i + slot_dim; ++j) {
                  y_data[offset + j] =
                      (x_data[offset + j] - means_data[j]) * scales_data[j];
                }
              }
            }

            offset += item_size;
          }
        } else {
338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387
          if (!ctx.Attr<bool>("enable_scale_and_shift") && slot_dim <= 0) {
            EigenArrayMap<T>(y_data, C, N) =
                (ConstEigenArrayMap<T>(x->data<T>(), C, N).colwise() -
                 means_arr)
                    .colwise() *
                scales_arr;
          } else if (ctx.Attr<bool>("enable_scale_and_shift") &&
                     slot_dim <= 0) {
            const auto *scale_w = ctx.Input<Tensor>("scale_w");
            const auto *bias = ctx.Input<Tensor>("bias");
            ConstEigenVectorArrayMap<T> scale_w_arr(scale_w->data<T>(), C);
            ConstEigenVectorArrayMap<T> bias_arr(bias->data<T>(), C);

            Eigen::Array<T, Eigen::Dynamic, 1> new_scale =
                scales_arr * scale_w_arr;
            Eigen::Array<T, Eigen::Dynamic, 1> new_bias =
                bias_arr - means_arr * scales_arr * scale_w_arr;
            EigenArrayMap<T>(y_data, C, N) =
                (ConstEigenArrayMap<T>(x->data<T>(), C, N).colwise() *
                 new_scale)
                    .colwise() +
                new_bias;

          } else {
            const int item_size = x->numel() / N;
            const auto *scale_w = ctx.Input<Tensor>("scale_w");
            const auto *bias = ctx.Input<Tensor>("bias");
            const T *scale_w_data = scale_w->data<T>();
            const T *bias_data = bias->data<T>();
            // location of show number in one embedding
            int offset = 0;
            for (int k = 0; k < N; ++k) {
              for (int i = 0; i < item_size; i += slot_dim) {
                if (x_data[offset + i] > -min_precision &&
                    x_data[offset + i] < min_precision) {
                  // show = 0
                  memset(y_data + offset + i, 0, sizeof(T) * slot_dim);
                } else {
                  for (int j = i; j < i + slot_dim; ++j) {
                    y_data[offset + j] = ((x_data[offset + j] - means_data[j]) *
                                          scales_data[j]) *
                                             scale_w_data[j] +
                                         bias_data[j];
                  }
                }
              }  // end for i

              offset += item_size;
            }  // end for k
          }
388
        }
H
heqiaozhi 已提交
389 390 391
        break;
      }
      default:
392
        PADDLE_THROW(platform::errors::InvalidArgument(
Y
yaoxuefeng 已提交
393
            "Unknown storage order: %d, please use NCHW or NHWC", data_layout));
H
heqiaozhi 已提交
394 395 396 397 398 399 400 401 402 403
    }
  }
};

class DataNormGradOp : public framework::OperatorWithKernel {
 public:
  using framework::OperatorWithKernel::OperatorWithKernel;

  void InferShape(framework::InferShapeContext *ctx) const override {
    // check input
404 405 406
    OP_INOUT_CHECK(ctx->HasInput("X"), "Input", "X", "DataNormGrad");
    OP_INOUT_CHECK(ctx->HasInput(framework::GradVarName("Y")), "Input",
                   framework::GradVarName("Y"), "DataNormGrad");
H
hutuxian 已提交
407 408 409 410 411 412 413 414 415 416 417 418
    PADDLE_ENFORCE_EQ(
        ctx->HasOutput("BatchSize"), true,
        platform::errors::NotFound(
            "Output(BatchSize) of DataNormGradOp should not be null."));
    PADDLE_ENFORCE_EQ(
        ctx->HasOutput("BatchSum"), true,
        platform::errors::NotFound(
            "Output(BatchSum) of DataNormGradOp should not be null."));
    PADDLE_ENFORCE_EQ(
        ctx->HasOutput("BatchSquareSum"), true,
        platform::errors::NotFound(
            "Output(BatchSquareSum) of DataNormGradOp should not be null."));
419 420
    OP_INOUT_CHECK(ctx->HasInput("Means"), "Input", "Means", "DataNormGrad");
    OP_INOUT_CHECK(ctx->HasInput("Scales"), "Input", "Scales", "DataNormGrad");
421 422
    bool enable_scale_and_shift =
        ctx->Attrs().Get<bool>("enable_scale_and_shift");
H
heqiaozhi 已提交
423
    // check output
424 425 426 427 428 429 430 431
    OP_INOUT_CHECK(ctx->HasOutput(framework::GradVarName("BatchSize")),
                   "Output", framework::GradVarName("BatchSize"),
                   "DataNormGrad");
    OP_INOUT_CHECK(ctx->HasOutput(framework::GradVarName("BatchSum")), "Output",
                   framework::GradVarName("BatchSum"), "DataNormGrad");
    OP_INOUT_CHECK(ctx->HasOutput(framework::GradVarName("BatchSquareSum")),
                   "Output", framework::GradVarName("BatchSquareSum"),
                   "DataNormGrad");
H
heqiaozhi 已提交
432 433 434 435 436 437 438 439

    const auto x_dims = ctx->GetInputDim("X");
    const DataLayout data_layout = framework::StringToDataLayout(
        ctx->Attrs().Get<std::string>("data_layout"));
    const int C =
        (data_layout == DataLayout::kNCHW ? x_dims[1]
                                          : x_dims[x_dims.size() - 1]);

440 441 442
    if (ctx->HasOutput(framework::GradVarName("X"))) {
      ctx->SetOutputDim(framework::GradVarName("X"), x_dims);
    }
H
heqiaozhi 已提交
443 444 445
    ctx->SetOutputDim(framework::GradVarName("BatchSize"), {C});
    ctx->SetOutputDim(framework::GradVarName("BatchSum"), {C});
    ctx->SetOutputDim(framework::GradVarName("BatchSquareSum"), {C});
446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461
    if (enable_scale_and_shift) {
      const bool has_scale_grad =
          ctx->HasOutput(framework::GradVarName("scale_w"));
      const bool has_bias_grad = ctx->HasOutput(framework::GradVarName("bias"));

      PADDLE_ENFORCE_EQ((has_scale_grad == has_bias_grad), true,
                        platform::errors::InvalidArgument(
                            "Output(Scale@GRAD) and Output(Bias@GRAD)"
                            "must be null or not be null at same time. "
                            "But now, has Scale@Grad=[%d], has Bias@GRAD=[%d]",
                            has_scale_grad, has_bias_grad));
      if (has_scale_grad) {
        ctx->SetOutputDim(framework::GradVarName("scale_w"), {C});
        ctx->SetOutputDim(framework::GradVarName("bias"), {C});
      }
    }
H
heqiaozhi 已提交
462 463 464 465 466 467 468
  }

 protected:
  framework::OpKernelType GetExpectedKernelType(
      const framework::ExecutionContext &ctx) const override {
    const auto *var = ctx.InputVar(framework::GradVarName("Y"));
    if (var == nullptr) {
469 470
      PADDLE_THROW(platform::errors::InvalidArgument(
          "Y@GRAD can not be found for computation"));
H
heqiaozhi 已提交
471 472 473 474 475 476 477 478
    }
    const Tensor *t = nullptr;
    if (var->IsType<Tensor>()) {
      t = &var->Get<Tensor>();
    } else if (var->IsType<LoDTensor>()) {
      t = &var->Get<LoDTensor>();
    }
    if (t == nullptr) {
479 480
      PADDLE_THROW(platform::errors::InvalidArgument(
          "Y@GRAD can not be found for computation"));
H
heqiaozhi 已提交
481 482 483 484 485 486
    }

    // TODO(pzelazko-intel): enable MKLDNN layout when it's ready
    framework::LibraryType library = framework::LibraryType::kPlain;
    framework::DataLayout layout = framework::DataLayout::kAnyLayout;

H
heqiaozhi 已提交
487 488
#ifdef PADDLE_WITH_MKLDNN
    if (library == framework::LibraryType::kPlain &&
489
        this->CanMKLDNNBeUsed(ctx)) {
H
heqiaozhi 已提交
490 491 492 493 494
      library = framework::LibraryType::kMKLDNN;
      layout = framework::DataLayout::kMKLDNN;
    }
#endif

495 496 497
    return framework::OpKernelType(
        OperatorWithKernel::IndicateVarDataType(ctx, "X"), ctx.GetPlace(),
        layout, library);
H
heqiaozhi 已提交
498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517
  }
};

template <typename T>
class DataNormGradKernel<platform::CPUDeviceContext, T>
    : public framework::OpKernel<T> {
 public:
  void Compute(const framework::ExecutionContext &ctx) const override {
    const auto *x = ctx.Input<Tensor>("X");
    const auto *d_y = ctx.Input<Tensor>(framework::GradVarName("Y"));
    const auto *scales = ctx.Input<Tensor>("Scales");
    const auto *means = ctx.Input<Tensor>("Means");

    const std::string data_layout_str = ctx.Attr<std::string>("data_layout");
    const DataLayout data_layout =
        framework::StringToDataLayout(data_layout_str);

    // Get the size for each dimension.
    // NCHW [batch_size, in_channels, in_height, in_width]
    const auto &x_dims = x->dims();
518 519
    PADDLE_ENFORCE_EQ(x_dims.size(), 2, platform::errors::InvalidArgument(
                                            "The Input dim size should be 2"));
H
heqiaozhi 已提交
520 521 522 523 524
    const int N = x_dims[0];
    const int C =
        (data_layout == DataLayout::kNCHW ? x_dims[1]
                                          : x_dims[x_dims.size() - 1]);
    // init output
525 526 527 528
    Tensor *d_x = nullptr;
    if (ctx.HasOutput(framework::GradVarName("X"))) {
      d_x = ctx.Output<Tensor>(framework::GradVarName("X"));
    }
529

H
heqiaozhi 已提交
530 531 532 533 534 535
    auto *d_batch_size =
        ctx.Output<Tensor>(framework::GradVarName("BatchSize"));
    auto *d_batch_sum = ctx.Output<Tensor>(framework::GradVarName("BatchSum"));
    auto *d_batch_square_sum =
        ctx.Output<Tensor>(framework::GradVarName("BatchSquareSum"));

536 537 538 539 540
    const T *mean_data = means->data<T>();
    const T *inv_var_data = scales->data<T>();
    ConstEigenVectorArrayMap<T> mean_arr(mean_data, C);
    ConstEigenVectorArrayMap<T> inv_var_arr(inv_var_data, C);

541 542 543 544 545 546 547
    T *d_batch_size_data = d_batch_size->mutable_data<T>(ctx.GetPlace());
    T *d_batch_sum_data = d_batch_sum->mutable_data<T>(ctx.GetPlace());
    T *d_batch_square_sum_data =
        d_batch_square_sum->mutable_data<T>(ctx.GetPlace());
    EigenVectorArrayMap<T> d_batch_size_arr(d_batch_size_data, C);
    EigenVectorArrayMap<T> d_batch_sum_arr(d_batch_sum_data, C);
    EigenVectorArrayMap<T> d_batch_square_sum_arr(d_batch_square_sum_data, C);
H
heqiaozhi 已提交
548 549 550
    d_batch_size_arr.setZero();
    d_batch_sum_arr.setZero();
    d_batch_square_sum_arr.setZero();
551 552
    const T *x_data = x->data<T>();
    const T *means_data = means->data<T>();
H
heqiaozhi 已提交
553 554

    const float epsilon = ctx.Attr<float>("epsilon");
555 556 557
    T min_precision = 1e-7f;
    const int slot_dim = ctx.Attr<int>("slot_dim");
    switch (data_layout) {  // it's two dimensions, make no difference
H
heqiaozhi 已提交
558 559 560 561 562 563
      case DataLayout::kNCHW:
      case DataLayout::kNHWC: {
        ConstEigenVectorArrayMap<T> scales_arr(scales->data<T>(), C);
        ConstEigenVectorArrayMap<T> means_arr(means->data<T>(), C);
        ConstEigenArrayMap<T> x_arr(x->data<T>(), C, N);
        ConstEigenArrayMap<T> d_y_arr(d_y->data<T>(), C, N);
564 565 566
        if (d_x != nullptr) {
          EigenArrayMap<T> d_x_arr(d_x->mutable_data<T>(ctx.GetPlace()), C, N);
          d_x_arr.setZero();
567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646
          if (!ctx.Attr<bool>("enable_scale_and_shift")) {
            for (int nc = 0; nc < N; ++nc) {
              d_x_arr.col(nc) = d_y_arr.col(nc) * scales_arr;
            }
          } else {
            const auto *scale_w = ctx.Input<Tensor>("scale_w");
            auto *d_scale =
                ctx.Output<Tensor>(framework::GradVarName("scale_w"));
            auto *d_bias = ctx.Output<Tensor>(framework::GradVarName("bias"));
            ConstEigenVectorArrayMap<T> scale_arr(scale_w->data<T>(), C);
            T *d_bias_data = nullptr;
            T *d_scale_data = nullptr;

            d_scale->mutable_data<T>(ctx.GetPlace());
            d_bias->mutable_data<T>(ctx.GetPlace());
            d_bias_data = d_bias->mutable_data<T>(ctx.GetPlace());
            d_scale_data = d_scale->mutable_data<T>(ctx.GetPlace());

            EigenVectorArrayMap<T> d_bias_arr(d_bias_data, C);
            EigenVectorArrayMap<T> d_scale_arr(d_scale_data, C);
            Tensor dy_sum;
            dy_sum.Resize({C});
            dy_sum.mutable_data<T>(ctx.GetPlace());
            EigenVectorArrayMap<T> dy_sum_arr(
                dy_sum.mutable_data<T>(ctx.GetPlace()), C);
            Tensor dy_mul_x_sub_mean_mul_invstd_sum;
            dy_mul_x_sub_mean_mul_invstd_sum.Resize({C});
            dy_mul_x_sub_mean_mul_invstd_sum.mutable_data<T>(ctx.GetPlace());
            EigenVectorArrayMap<T> dy_mul_x_sub_mean_mul_invstd_sum_arr(
                dy_mul_x_sub_mean_mul_invstd_sum.mutable_data<T>(
                    ctx.GetPlace()),
                C);

            dy_sum_arr.setZero();
            dy_mul_x_sub_mean_mul_invstd_sum_arr.setZero();

            if (slot_dim <= 0) {
              for (int n = 0; n < N; ++n) {
                dy_sum_arr += d_y_arr.col(n);
                dy_mul_x_sub_mean_mul_invstd_sum_arr +=
                    ((x_arr.col(n) - mean_arr) * inv_var_arr * d_y_arr.col(n));
              }
              if (d_scale && d_bias) {
                d_bias_arr = dy_sum_arr;
                d_scale_arr = dy_mul_x_sub_mean_mul_invstd_sum_arr;
              }
              for (int nc = 0; nc < N; ++nc) {
                d_x_arr.col(nc) = d_y_arr.col(nc) * scales_arr * scale_arr;
              }
            } else {
              int offset = 0;
              const int item_size = x->numel() / N;
              T *d_x_data = d_x->mutable_data<T>(ctx.GetPlace());
              T *d_scale_data = d_scale->mutable_data<T>(ctx.GetPlace());
              T *d_bias_data = d_bias->mutable_data<T>(ctx.GetPlace());
              const T *dy_data = d_y->data<T>();
              const T *scales_data = scales->data<T>();
              const T *scale_w_data = scale_w->data<T>();
              const T *x_data = x->data<T>();
              for (int i = 0; i < item_size; i++) {
                d_bias_data[i] = 0;
                d_scale_data[i] = 0;
              }
              for (int k = 0; k < N; ++k) {
                for (int i = 0; i < item_size; i += slot_dim) {
                  if (!(x_data[offset + i] > -min_precision &&
                        x_data[offset + i] < min_precision)) {
                    // show != 0
                    for (int j = i; j < i + slot_dim; ++j) {
                      d_x_data[offset + j] = dy_data[offset + j] *
                                             scales_data[j] * scale_w_data[j];
                      d_bias_data[j] += dy_data[offset + j];
                      d_scale_data[j] += (x_data[offset + j] - mean_data[j]) *
                                         inv_var_data[j] * dy_data[offset + j];
                    }
                  }
                }
                offset += item_size;
              }
            }
647
          }
H
heqiaozhi 已提交
648 649
        }

650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698
        if (slot_dim > 0 && N > 0) {
          // if slot_dim is set and batch size is larger than zero, we choose
          // to check if show number is zero, if so, skip update statistics.
          int offset = 0;
          const int item_size = x->numel() / N;
          for (int k = 0; k < N; ++k) {
            for (int i = 0; i < item_size; i += slot_dim) {
              if (!(x_data[offset + i] > -min_precision &&
                    x_data[offset + i] < min_precision)) {
                // show != 0
                for (int j = i; j < i + slot_dim; ++j) {
                  d_batch_size_data[j] += 1;
                  d_batch_sum_data[j] += x_data[offset + j];
                  d_batch_square_sum_data[j] +=
                      (x_data[offset + j] - means_data[j]) *
                      (x_data[offset + j] - means_data[j]);
                }
              }
            }
            offset += item_size;
          }

          for (int i = 0; i < item_size; i += slot_dim) {
            for (int j = i; j < i + slot_dim; ++j) {
              if (d_batch_size_data[j] >= 1) {
                d_batch_sum_data[j] /= d_batch_size_data[j];
                d_batch_square_sum_data[j] =
                    d_batch_square_sum_data[j] / d_batch_size_data[j] +
                    d_batch_size_data[j] * epsilon;
                d_batch_size_data[j] = 1;
              }
            }
          }
        } else {
          // calculate data sum and squre sum
          Eigen::Array<T, Eigen::Dynamic, 1> sample_sum(C);
          Eigen::Array<T, Eigen::Dynamic, 1> sample_square_sum(C);
          // calculate data sample sum and square sum
          sample_sum.setZero();
          sample_square_sum.setZero();
          for (int nc = 0; nc < N; ++nc) {
            sample_sum += x_arr.col(nc);
            sample_square_sum += (x_arr.col(nc) - means_arr).square();
          }
          // calculate gradient
          d_batch_size_arr.setConstant(N);
          d_batch_sum_arr = sample_sum;
          d_batch_square_sum_arr =
              sample_square_sum + d_batch_size_arr * epsilon;
H
heqiaozhi 已提交
699 700 701 702
        }
        break;
      }
      default:
703
        PADDLE_THROW(platform::errors::InvalidArgument(
Y
yaoxuefeng 已提交
704 705
            "Unknown storage order: %s, please use NCHW or NHWC",
            data_layout_str));
H
heqiaozhi 已提交
706 707 708 709
    }
  }
};

H
hong 已提交
710 711
template <typename T>
class DataNormGradMaker : public framework::SingleGradOpMaker<T> {
H
heqiaozhi 已提交
712
 public:
H
hong 已提交
713
  using framework::SingleGradOpMaker<T>::SingleGradOpMaker;
H
heqiaozhi 已提交
714 715

 protected:
716
  void Apply(GradOpPtr<T> op) const override {
H
heqiaozhi 已提交
717
    op->SetType("data_norm_grad");
H
hong 已提交
718 719 720
    op->SetInput("X", this->Input("X"));
    op->SetInput(framework::GradVarName("Y"), this->OutputGrad("Y"));

721 722
    op->SetInput("scale_w", this->Input("scale_w"));
    op->SetInput("bias", this->Input("bias"));
H
hutuxian 已提交
723 724 725
    op->SetOutput("BatchSize", this->Input("BatchSize"));
    op->SetOutput("BatchSum", this->Input("BatchSum"));
    op->SetOutput("BatchSquareSum", this->Input("BatchSquareSum"));
H
hong 已提交
726 727 728 729 730 731 732 733 734 735
    op->SetInput("Scales", this->Output("Scales"));
    op->SetInput("Means", this->Output("Means"));

    op->SetAttrMap(this->Attrs());

    op->SetOutput(framework::GradVarName("X"), this->InputGrad("X"));
    op->SetOutput(framework::GradVarName("BatchSize"),
                  this->InputGrad("BatchSize"));
    op->SetOutput(framework::GradVarName("BatchSum"),
                  this->InputGrad("BatchSum"));
H
heqiaozhi 已提交
736
    op->SetOutput(framework::GradVarName("BatchSquareSum"),
H
hong 已提交
737
                  this->InputGrad("BatchSquareSum"));
738 739 740
    op->SetOutput(framework::GradVarName("scale_w"),
                  this->InputGrad("scale_w"));
    op->SetOutput(framework::GradVarName("bias"), this->InputGrad("bias"));
H
heqiaozhi 已提交
741 742 743 744 745 746 747 748
  }
};

}  // namespace operators
}  // namespace paddle

namespace ops = paddle::operators;
REGISTER_OPERATOR(data_norm, ops::DataNormOp, ops::DataNormOpMaker,
H
hong 已提交
749 750
                  ops::DataNormGradMaker<paddle::framework::OpDesc>,
                  ops::DataNormGradMaker<paddle::imperative::OpBase>);
H
heqiaozhi 已提交
751 752 753 754 755 756 757 758 759
REGISTER_OPERATOR(data_norm_grad, ops::DataNormGradOp);

REGISTER_OP_CPU_KERNEL(
    data_norm, ops::DataNormKernel<paddle::platform::CPUDeviceContext, float>,
    ops::DataNormKernel<paddle::platform::CPUDeviceContext, double>);
REGISTER_OP_CPU_KERNEL(
    data_norm_grad,
    ops::DataNormGradKernel<paddle::platform::CPUDeviceContext, float>,
    ops::DataNormGradKernel<paddle::platform::CPUDeviceContext, double>);
760 761 762 763 764 765 766
REGISTER_OP_VERSION(data_norm)
    .AddCheckpoint(
        R"ROC(
              upgrad data_norm op by adding scale_w to support scale and shift.)ROC",
        paddle::framework::compatible::OpVersionDesc().NewInput(
            "scale_w",
            "scale_w is used to do scale duirng data_norm like batchnorm "));