sample_logits_op.h 12.3 KB
Newer Older
X
xuezhong 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
/* 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. */

#pragma once

17
#include <unordered_set>
X
xuezhong 已提交
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
#include <vector>
#include "paddle/fluid/framework/eigen.h"
#include "paddle/fluid/framework/op_registry.h"
#include "paddle/fluid/framework/tensor_util.h"
#include "paddle/fluid/operators/math/math_function.h"
#include "paddle/fluid/operators/math/sample_prob.h"
#include "paddle/fluid/operators/math/softmax.h"

namespace paddle {
namespace operators {

using Tensor = framework::Tensor;
template <typename T, int MajorType = Eigen::RowMajor,
          typename IndexType = Eigen::DenseIndex>
using EigenMatrix = framework::EigenMatrix<T, MajorType, IndexType>;

template <typename T>
struct TolerableValue {
  HOSTDEVICE T operator()(const T& x) const {
37 38
    PADDLE_ENFORCE(std::is_floating_point<T>::value,
                   "TolerableValue should be float in sample_logits_op.");
X
xuezhong 已提交
39 40 41 42 43 44 45 46 47 48 49 50 51
    const T kApproInf = 1e20;
    if (x == INFINITY) return kApproInf;
    if (x == -INFINITY) return -kApproInf;
    return x;
  }
};

// UNDERSTAND: something like take_along_axis in numpy.
template <typename T>
static void CPUTakeAlongD1(const platform::DeviceContext& ctx,
                           const framework::Tensor& array,
                           const framework::Tensor& index,
                           framework::Tensor* value) {
52 53 54
  PADDLE_ENFORCE_EQ(
      platform::is_cpu_place(ctx.GetPlace()), true,
      platform::errors::InvalidArgument("This kernel only runs on CPU."));
X
xuezhong 已提交
55 56 57 58 59
  // UNDERSTAND: check shape src(B, C), index(B, K), out should also be (B, K)
  const auto batch_size = index.dims()[0];
  const auto num_take = index.dims()[1];
  const auto array_dims = array.dims();
  const auto idx_dims = index.dims();
60 61
  PADDLE_ENFORCE_EQ(idx_dims.size(), 2,
                    platform::errors::InvalidArgument(
62
                        "index of CPUTakeAlongD1 should be 2D. "
63 64 65 66
                        "But received shape = [%s] and dimension is %d.",
                        idx_dims, idx_dims.size()));
  PADDLE_ENFORCE_EQ(array_dims.size(), 2,
                    platform::errors::InvalidArgument(
67
                        "array of CPUTakeAlongD1 should be 2D. "
68 69 70 71
                        "But received shape = [%s] and dimension is %d.",
                        array_dims, array_dims.size()));
  PADDLE_ENFORCE_EQ(idx_dims[0], array_dims[0],
                    platform::errors::InvalidArgument(
72 73 74 75 76
                        "The first dimension of index and array of "
                        "CPUTakeAlongD1 should be equal. "
                        "But received index shape = [%s], array shape = [%s], "
                        "and the first dimensions are %d and %d.",
                        idx_dims, array_dims, idx_dims[0], array_dims[0]));
77 78 79
  PADDLE_ENFORCE_EQ(
      idx_dims, value->dims(),
      platform::errors::InvalidArgument(
80 81 82
          "index and array of CPUTakeAlongD1 should have the same shape. "
          "But received index shape = [%s], array shape = [%s].",
          idx_dims, value->dims()));
X
xuezhong 已提交
83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111

  // UNDERSTAND: no allocations here
  const T* p_array = array.data<T>();
  const int64_t* p_index = index.data<int64_t>();
  T* p_value = value->data<T>();

  // src slice size
  const auto array_slice_size = array_dims[1];

  // index slice size
  const auto idx_slice_size = idx_dims[1];
  const auto value_slice_size = idx_slice_size;

  for (int i = 0; i < batch_size; ++i) {
    for (int j = 0; j < num_take; ++j) {
      auto array_index = p_index[i * idx_slice_size + j];
      p_value[i * value_slice_size + j] =
          p_array[i * array_slice_size + array_index];
    }
  }
}

// UNDERSTAND: something like put_along_axis in numpy but if there is duplicate
// indices, scatter is done in += way.
template <typename T>
static void CPUPutAlongD1(const platform::DeviceContext& ctx,
                          framework::Tensor* array,
                          const framework::Tensor& index,
                          const framework::Tensor& value) {
112 113 114
  PADDLE_ENFORCE_EQ(
      platform::is_cpu_place(ctx.GetPlace()), true,
      platform::errors::InvalidArgument("This kernel only runs on CPU."));
X
xuezhong 已提交
115 116 117 118 119
  // UNDERSTAND: check shape src(B, C), index(B, K), out should also be (B, K)
  const auto batch_size = index.dims()[0];
  const auto num_put = index.dims()[1];
  auto array_dims = array->dims();
  auto idx_dims = index.dims();
120 121
  PADDLE_ENFORCE_EQ(idx_dims.size(), 2,
                    platform::errors::InvalidArgument(
122
                        "index of CPUPutAlongD1 should be 2D. "
123 124 125 126
                        "But received shape = [%s] and dimension is %d.",
                        idx_dims, idx_dims.size()));
  PADDLE_ENFORCE_EQ(array_dims.size(), 2,
                    platform::errors::InvalidArgument(
127
                        "array of CPUPutAlongD1 should be 2D. "
128 129 130 131
                        "But received shape = [%s] and dimension is %d.",
                        array_dims, array_dims.size()));
  PADDLE_ENFORCE_EQ(idx_dims[0], array_dims[0],
                    platform::errors::InvalidArgument(
132 133 134 135 136
                        "The first dimension of index and array of "
                        "CPUPutAlongD1 should be equal. "
                        "But received index shape = [%s], array shape = [%s], "
                        "and the first dimensions are %d and %d.",
                        idx_dims, array_dims, idx_dims[0], array_dims[0]));
137 138 139
  PADDLE_ENFORCE_EQ(
      idx_dims, value.dims(),
      platform::errors::InvalidArgument(
140 141 142
          "index and array of CPUPutAlongD1 should have the same shape. "
          "But received index shape = [%s], array shape = [%s].",
          idx_dims, value.dims()));
X
xuezhong 已提交
143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192

  // UNDERSTAND: no allocations here
  T* p_array = array->data<T>();
  const int64_t* p_index = index.data<int64_t>();
  const T* p_value = value.data<T>();

  // slice sizes
  const auto array_slice_size = array_dims[1];
  const auto idx_slice_size = idx_dims[1];
  const auto value_slice_size = idx_slice_size;

  for (int i = 0; i < batch_size; ++i) {
    for (int j = 0; j < num_put; ++j) {
      auto array_index = p_index[i * idx_slice_size + j];
      p_array[i * array_slice_size + array_index] +=
          p_value[i * value_slice_size + j];
    }
  }
}

// UNDERSTAND: compute accidentdal hits from samples and minus corresponding
// logits by a float max, here 1e20
template <typename T>
static void compute_remove_accidental_hits(const platform::DeviceContext& ctx,
                                           framework::Tensor* sampled_logits,
                                           const framework::Tensor& samples,
                                           const int num_true) {
  const auto batch_size = sampled_logits->dims()[0];
  const auto num_sampled_classes = sampled_logits->dims()[1];
  T* sampled_logits_data = sampled_logits->data<T>();
  const auto samples_data = samples.data<int64_t>();

  std::unordered_set<int64_t> tmp_true_labels;
  for (int i = 0; i < batch_size; ++i) {
    tmp_true_labels.clear();
    tmp_true_labels.insert(samples_data + i * num_sampled_classes,
                           samples_data + i * num_sampled_classes + num_true);
    for (int j = num_true; j < num_sampled_classes; ++j) {
      const auto idx = i * num_sampled_classes + j;
      if (tmp_true_labels.find(samples_data[idx]) != tmp_true_labels.end())
        sampled_logits_data[idx] -= 1e20;
    }
  }
}

template <typename T>
class SampleLogitsKernel : public framework::OpKernel<T> {
 public:
  using Tensor = framework::Tensor;
  void Compute(const framework::ExecutionContext& context) const override {
193 194 195
    PADDLE_ENFORCE_EQ(
        platform::is_cpu_place(context.GetPlace()), true,
        platform::errors::InvalidArgument("this kernel only runs on cpu."));
X
xuezhong 已提交
196 197 198
    VLOG(3) << "Enter SampleLogitsKernel";
    // get necessary inputs
    const Tensor* logits = context.Input<Tensor>("Logits");
X
xuezhong 已提交
199
    const Tensor* labels = context.Input<Tensor>("Labels");
X
xuezhong 已提交
200 201 202 203 204

    // get necessary outputs
    Tensor* samples = context.Output<Tensor>("Samples");
    Tensor* probabilities = context.Output<Tensor>("Probabilities");
    Tensor* sampled_logits = context.Output<Tensor>("SampledLogits");
X
xuezhong 已提交
205
    Tensor* sampled_labels = context.Output<Tensor>("SampledLabels");
X
xuezhong 已提交
206 207 208 209

    // shapes
    const auto batch_size = logits->dims()[0];
    const auto num_classes = logits->dims()[1];
X
xuezhong 已提交
210 211
    const auto labels_dim = labels->dims();
    const auto num_true = labels_dim[1];
X
xuezhong 已提交
212 213 214 215
    const auto samples_dim = samples->dims();

    // attrs
    const auto num_samples = context.Attr<int>("num_samples");
X
xuezhong 已提交
216 217
    const bool use_customized_samples =
        context.Attr<bool>("use_customized_samples");
X
xuezhong 已提交
218 219 220 221 222 223 224 225 226
    const bool remove_accidental_hits =
        context.Attr<bool>("remove_accidental_hits");

    // device contexts
    auto& dev_ctx =
        context.template device_context<platform::CPUDeviceContext>();

    // UNDERSTAND: allocate memories for temporaries
    sampled_logits->mutable_data<T>(samples_dim, context.GetPlace());
X
xuezhong 已提交
227 228 229 230 231 232 233 234 235 236 237 238 239
    auto sampled_labels_data =
        sampled_labels->mutable_data<int64_t>(labels_dim, context.GetPlace());
    for (int i = 0; i < batch_size; ++i) {
      for (int j = 0; j < num_true; ++j) {
        sampled_labels_data[i * num_true + j] = j;
      }
    }

    if (use_customized_samples) {
      const Tensor* customized_samples =
          context.Input<Tensor>("CustomizedSamples");
      const Tensor* customized_probabilities =
          context.Input<Tensor>("CustomizedProbabilities");
240 241 242 243 244 245 246 247 248
      PADDLE_ENFORCE_EQ(customized_samples, samples,
                        platform::errors::InvalidArgument(
                            "CustomizedSamples must be the same Tensor with "
                            "Samples when use_customized_samples = True"));
      PADDLE_ENFORCE_EQ(
          customized_probabilities, probabilities,
          platform::errors::InvalidArgument(
              "CustomizedProbabilities must be the same Tensor with "
              "Probabilities when use_customized_samples = True"));
X
xuezhong 已提交
249 250 251 252 253 254 255 256
    } else {
      samples->mutable_data<int64_t>(context.GetPlace());
      probabilities->mutable_data<T>(samples_dim, context.GetPlace());
      // UNDERSTAND: sampling
      const auto seed = context.Attr<int>("seed");
      auto sampler_with_prob =
          math::SampleWithProb<platform::CPUDeviceContext, T>();
      sampler_with_prob(dev_ctx, math::LogUniformSampler(num_classes, seed),
X
xuezhong 已提交
257
                        num_samples, labels, samples, probabilities);
X
xuezhong 已提交
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298
    }

    // UNDERSTAND: gather sampled logits and remove accidental hits if needed
    CPUTakeAlongD1<T>(dev_ctx, *logits, *samples, sampled_logits);
    if (remove_accidental_hits) {
      compute_remove_accidental_hits<T>(dev_ctx, sampled_logits, *samples,
                                        num_true);
    }

    // subtracted sampled logits with logQ(y|x)
    auto probs = EigenMatrix<T>::From(*probabilities);
    auto smp_logits = EigenMatrix<T>::From(*sampled_logits);
    smp_logits.device(*dev_ctx.eigen_device()) =
        (smp_logits - probs.log().unaryExpr(TolerableValue<T>()))
            .unaryExpr(TolerableValue<T>());
  }
};

template <typename T>
class SampleLogitsGradKernel : public framework::OpKernel<T> {
 public:
  using Tensor = framework::Tensor;
  void Compute(const framework::ExecutionContext& context) const override {
    auto logits_grad = context.Output<Tensor>(framework::GradVarName("Logits"));
    const Tensor* samples = context.Input<Tensor>("Samples");
    const Tensor* sampled_logits_grad =
        context.Input<Tensor>(framework::GradVarName("SampledLogits"));
    logits_grad->mutable_data<T>(context.GetPlace());

    auto& dev_ctx =
        context.template device_context<platform::CPUDeviceContext>();
    math::SetConstant<platform::CPUDeviceContext, T> set_zero;
    set_zero(dev_ctx, logits_grad, static_cast<T>(0));

    // UNDERSTAND: scatter it back to logit_grad
    CPUPutAlongD1<T>(dev_ctx, logits_grad, *samples, *sampled_logits_grad);
  }
};

}  // namespace operators
}  // namespace paddle