test=develop, remove sparse bias and add prefetch and related tests

paddle/fluid/operators/distributed/parameter_prefetch.cc

@@ -32,7 +32,7 @@ namespace paddle {
 namespace operators {
 namespace distributed {
 
-using Tensor = framework::Tensor;
+using LoDTensor = framework::LoDTensor;
 using LoDTensor = framework::LoDTensor;
 using SelectedRows = framework::SelectedRows;
 using DDim = framework::DDim;
@@ -120,8 +120,8 @@ static void MergeMultipleVarsIntoOneBySection(
 
   PADDLE_ENFORCE_GT(
       out_tensor->numel(), 0,
-      "When calling this method, the Tensor's numel must larger than zero. "
-      "Please check Tensor::Resize has been called first.");
+      "When calling this method, the LoDTensor's numel must larger than zero. "
+      "Please check LoDTensor::Resize has been called first.");
 
   auto* out_tensor_data = out_tensor->mutable_data<float>(id_tensor.place());
 
@@ -144,7 +144,7 @@ static void MergeMultipleVarsIntoOneBySection(
 
       auto row_numel = dims[1];
 
-      for (size_t i = 0; i < dims[0]; ++i) {
+      for (int64_t i = 0; i < dims[0]; ++i) {
         auto id = ids_in_this_section[i];
         auto origin_id = id + abs_sections[section_idx];
         auto& offsets = id_to_offset[origin_id];
@@ -201,7 +201,7 @@ void prefetch(const std::string& id_name, const std::string& out_name,
   std::vector<int64_t> ids_vector;
   if (platform::is_cpu_place(id_tensor.place())) {
     auto* id_data = id_tensor.data<int64_t>();
-    for (size_t i = 0; i < id_tensor.numel(); ++i) {
+    for (int64_t i = 0; i < id_tensor.numel(); ++i) {
       ids_vector.push_back(id_data[i]);
     }
   } else {
@@ -209,7 +209,7 @@ void prefetch(const std::string& id_name, const std::string& out_name,
     PADDLE_THROW("paddle is not compiled with CUDA!");
 #else
     auto cpu_place = platform::CPUPlace();
-    framework::Tensor cpu_tensor;
+    framework::LoDTensor cpu_tensor;
     auto* cpu_tensor_data =
         cpu_tensor.mutable_data<int64_t>(id_tensor.dims(), cpu_place);
     auto stream =

paddle/fluid/operators/distributed/parameter_prefetch.h

@@ -30,6 +30,30 @@ void prefetch(const std::string& id_name, const std::string& out_name,
               const framework::ExecutionContext& context,
               const framework::Scope& scope);
 
+template <typename T>
+void prefetch_with_reconstruct(const std::string& id_name,
+                               const std::string& out_name,
+                               const std::vector<std::string>& table_names,
+                               const std::vector<std::string>& epmap,
+                               const std::vector<int>& height_sections,
+                               const framework::ExecutionContext& context,
+                               const framework::Scope& scope,
+                               framework::LoDTensor* original) {
+  prefetch(id_name, out_name, table_names, epmap, height_sections, context,
+           scope);
+  auto& out = scope.FindVar(out_name)->Get<framework::LoDTensor>();
+  auto& ids = scope.FindVar(id_name)->Get<framework::LoDTensor>();
+  auto* original_value = original->data<T>();
+  auto* out_value = out.data<T>();
+  size_t original_width = original->numel() / original->dims()[0];
+
+  for (int64_t i = 0; i < ids.numel(); i++) {
+    const T* out_rows = out_value + original_width * i;
+    T* original_row = original_value + original_width * ids.data<int64_t>()[i];
+    std::memcpy(original_row, out_rows, original_width * sizeof(T));
+  }
+}
+
 };  // namespace distributed
 };  // namespace operators
 };  // namespace paddle

paddle/fluid/operators/hierarchical_sigmoid_op.cc

@@ -67,6 +67,11 @@ class HierarchicalSigmoidOp : public framework::OperatorWithKernel {
     PADDLE_ENFORCE(ctx->HasOutput("Out"), "Output(Out) should not be null.");
     PADDLE_ENFORCE(ctx->HasOutput("PreOut"),
                    "Output(PreOut) should not be null.");
+    auto with_prefetch = ctx->Attrs().Get<bool>("remote_prefetch");
+    if (with_prefetch) {
+      PADDLE_ENFORCE(ctx->HasOutput("W_Out"),
+                     "Output(W_Out) should not be null.");
+    }
     const int64_t batch_size = ctx->GetInputDim("X")[0];
     std::vector<int64_t> output_shape({batch_size, 1});
     ctx->SetOutputDim("Out", framework::make_ddim(output_shape));
@@ -96,7 +101,7 @@ class HierarchicalSigmoidOpMaker : public framework::OpProtoAndCheckerMaker {
     AddInput("Label",
              "(LoDTensor, required), The labels of training data. It's a"
              "tensor with shape [N, 1].");
-    AddInput("PTable",
+    AddInput("PathTable",
              "(LoDTensor, optional), The Path Table from root to current word"
              "it should have shape like [N, L], L is the length of the Path")
         .AsDispensable();
@@ -120,8 +125,30 @@ class HierarchicalSigmoidOpMaker : public framework::OpProtoAndCheckerMaker {
               "[batch_size, code_length], where code_length represents the "
               "maximum path length from root to leaf nodes.")
         .AsIntermediate();
+    AddOutput(
+        "W_Out",
+        "(LoDTensor, optinal) using input 'W' as Output to make it mutable"
+        "When we are using prefetch")
+        .AsIntermediate();
     AddAttr<AttrType>("num_classes", "(int, optional), The number of classes")
         .SetDefault(2);
+    // for parameter prefetch
+    AddAttr<bool>("remote_prefetch", "").SetDefault(false);
+    AddAttr<int>("trainer_id", "trainer id from 0 ~ worker_num.").SetDefault(0);
+    AddAttr<std::vector<int>>("height_sections",
+                              "Height for each output SelectedRows.")
+        .SetDefault(std::vector<int>({}));
+    AddAttr<std::vector<std::string>>(
+        "epmap",
+        "(string vector, default 127.0.0.1:6164)"
+        "Server endpoints in the order of input variables for mapping")
+        .SetDefault({});
+    AddAttr<std::vector<std::string>>(
+        "table_names",
+        "(string vector, the splited table names that will be fetched from "
+        "parameter server)"
+        "in the order of input variables for mapping")
+        .SetDefault({});
     AddComment(R"DOC(
 The hierarchical sigmoid operator organize the classes into a binary tree.
 At each node, a sigmoid function is used to calculate the probability of
@@ -191,24 +218,18 @@ class HierarchicalSigmoidGradOpGradVarTypeInference
                << " is set to SelectedRows";
       block->Var(w_grad_var_name)
           ->SetType(framework::proto::VarType::SELECTED_ROWS);
-      if (hasBias) {
-        VLOG(30) << "hierarchical_sigmoid_grad op "
-                 << framework::GradVarName("Bias") << " is set to SelectedRows";
-        block->Var(bias_grad_var_name)
-            ->SetType(framework::proto::VarType::SELECTED_ROWS);
-      }
     } else {
       VLOG(30) << "hierarchical_sigmoid_grad op " << framework::GradVarName("W")
                << " is set to LoDTensor";
       block->Var(w_grad_var_name)
           ->SetType(framework::proto::VarType::LOD_TENSOR);
+    }
     if (hasBias) {
       VLOG(30) << "hierarchical_sigmoid_grad op "
                << framework::GradVarName("Bias") << " is set to LoDTensor";
       block->Var(bias_grad_var_name)
           ->SetType(framework::proto::VarType::LOD_TENSOR);
     }
-    }
     block->Var(w_grad_var_name)->SetDataType(block->Var("W")->GetDataType());
   }
 };

paddle/fluid/operators/hierarchical_sigmoid_op.h

@@ -14,7 +14,9 @@ limitations under the License. */
 
 #pragma once
 #include <iostream>
+#include <iterator>
 #include <set>
+#include <string>
 #include <vector>
 #include "paddle/fluid/framework/mixed_vector.h"
 #include "paddle/fluid/framework/op_registry.h"
@@ -24,6 +26,10 @@ limitations under the License. */
 #include "paddle/fluid/operators/math/matrix_bit_code.h"
 #include "paddle/fluid/platform/transform.h"
 
+#ifdef PADDLE_WITH_DISTRIBUTE
+#include "paddle/fluid/operators/distributed/parameter_prefetch.h"
+#endif
+
 namespace paddle {
 namespace operators {
 
@@ -49,13 +55,55 @@ class HierarchicalSigmoidOpKernel : public framework::OpKernel<T> {
   void Compute(const framework::ExecutionContext& ctx) const override {
     auto& in = detail::Ref(ctx.Input<framework::LoDTensor>("X"));
     auto& w = detail::Ref(ctx.Input<framework::LoDTensor>("W"));
-    auto* path = ctx.Input<framework::LoDTensor>("PTable");
+    auto* path = ctx.Input<framework::LoDTensor>("PathTable");
     auto* code = ctx.Input<framework::LoDTensor>("PathCode");
     auto& label = detail::Ref(ctx.Input<framework::LoDTensor>("Label"));
     auto* bias = ctx.Input<framework::LoDTensor>("Bias");
     auto* out = ctx.Output<framework::LoDTensor>("Out");
     auto* pre_out = ctx.Output<framework::LoDTensor>("PreOut");
     size_t num_classes = static_cast<size_t>(ctx.Attr<int>("num_classes"));
+    // for remote prefetch
+
+    auto epmap = ctx.Attr<std::vector<std::string>>("epmap");
+    if (!epmap.empty()) {
+      // if epmap is not empty, then the parameter will be fetched from remote
+      // parameter
+      // server
+      auto height_sections = ctx.Attr<std::vector<int>>("height_sections");
+      auto table_names = ctx.Attr<std::vector<std::string>>("table_names");
+      VLOG(3) << "path type is " << path->type().name();
+      std::vector<int64_t> real_rows = PathToRows(*path);
+      framework::Scope& local_scope = ctx.scope().NewScope();
+      auto* ids = local_scope.Var("Ids@Prefetch");
+      auto* x_tensor = ids->GetMutable<framework::LoDTensor>();
+
+      x_tensor->mutable_data<int64_t>(
+          framework::make_ddim({static_cast<int64_t>(real_rows.size()), 1}),
+          ctx.GetPlace());
+      // copy.
+
+      std::memcpy(x_tensor->data<int64_t>(), real_rows.data(),
+                  real_rows.size() * sizeof(int64_t));
+
+      framework::DDim w_dims = ctx.Input<Tensor>("W")->dims();
+      w_dims[0] = x_tensor->dims()[0];
+      auto* w_tensor =
+          local_scope.Var("W@Prefetch")->GetMutable<framework::LoDTensor>();
+      w_tensor->Resize(w_dims);
+
+#ifdef PADDLE_WITH_DISTRIBUTE
+      // w_Out is set to used by prefetch, never change it in other cases
+      auto* w_out = ctx.Output<framework::LoDTensor>("W_Out");
+      operators::distributed::prefetch_with_reconstruct<T>(
+          "Ids@Prefetch", "W@Prefetch", table_names, epmap, height_sections,
+          ctx, local_scope, w_out);
+#else
+      PADDLE_THROW(
+          "paddle is not compiled with distribute support, can not do "
+          "parameter prefetch!");
+#endif
+    }
+
     bool is_custom = false;
     if (path) {
       is_custom = true;
@@ -116,9 +164,8 @@ class HierarchicalSigmoidGradOpKernel : public framework::OpKernel<T> {
   void Compute(const framework::ExecutionContext& ctx) const override {
     auto& in = detail::Ref(ctx.Input<framework::LoDTensor>("X"));
     auto& w = detail::Ref(ctx.Input<framework::LoDTensor>("W"));
-    auto* path = ctx.Input<framework::LoDTensor>("PTable");
+    auto* path = ctx.Input<framework::LoDTensor>("PathTable");
     auto* code = ctx.Input<framework::LoDTensor>("PathCode");
-    auto* bias = ctx.Input<framework::LoDTensor>("Bias");
     auto* in_grad =
         ctx.Output<framework::LoDTensor>(framework::GradVarName("X"));
     bool is_sparse = ctx.Attr<bool>("is_sparse");
@@ -165,8 +212,6 @@ class HierarchicalSigmoidGradOpKernel : public framework::OpKernel<T> {
         pre_out_grad_mat * out_grad_mat.broadcast(bcast);
     // TODO(guosheng): multiply pre_out_grad with subgradient of clipping to
     // be consistent with the clipping in forward.
-
-    if (!is_sparse) {
     auto* bias_grad =
         ctx.Output<framework::LoDTensor>(framework::GradVarName("Bias"));
     if (bias_grad) {
@@ -174,6 +219,7 @@ class HierarchicalSigmoidGradOpKernel : public framework::OpKernel<T> {
       zero(dev_ctx, bias_grad, static_cast<T>(0.0));
       bit_code->AddGrad(pre_out_grad, bias_grad);
     }
+    if (!is_sparse) {
       auto* w_grad =
           ctx.Output<framework::LoDTensor>(framework::GradVarName("W"));
       w_grad->mutable_data<T>(ctx.GetPlace());
@@ -192,21 +238,6 @@ class HierarchicalSigmoidGradOpKernel : public framework::OpKernel<T> {
 
       w_grad_value->mutable_data<T>(temp_dim, ctx.GetPlace());
       zero(dev_ctx, w_grad_value, static_cast<T>(0.0));
-      auto* bias_grad =
-          ctx.Output<framework::SelectedRows>(framework::GradVarName("Bias"));
-      if (bias_grad) {
-        bias_grad->set_rows(real_rows);
-        // build ids -> rows index map
-        bias_grad->SyncIndex();
-        bias_grad->set_height(bias->dims()[0]);
-        auto* bias_grad_value = bias_grad->mutable_value();
-        std::vector<int64_t> dims = {static_cast<int64_t>(real_rows.size()),
-                                     bias->dims()[1]};
-        bias_grad_value->mutable_data<T>(framework::make_ddim(dims),
-                                         ctx.GetPlace());
-        zero(dev_ctx, bias_grad_value, static_cast<T>(0.0));
-        bit_code->AddGrad(pre_out_grad, bias_grad);
-      }
       bit_code->MulGradWeight(pre_out_grad, w_grad, in);
     }
     bit_code->MulGradError(pre_out_grad, w, in_grad);

paddle/fluid/operators/math/matrix_bit_code.cc

@@ -48,23 +48,6 @@ void MatrixBitCodeFunctor<T>::AddGrad(const framework::Tensor& tmat,
   }
 }
 
-template <typename T>
-void MatrixBitCodeFunctor<T>::AddGrad(const framework::Tensor& tmat,
-                                      framework::SelectedRows* vec) {
-  size_t batch_size = tmat.dims()[0];
-  size_t width = tmat.dims()[1];
-  for (size_t i = 0; i < batch_size; ++i) {
-    auto code = code_table_->get_code(i);
-    int code_length = code->get_length();
-    for (int j = 0; j < code_length; ++j) {
-      size_t index = code->calc_index(j);
-      int64_t row_index = vec->GetIndexFromId(static_cast<int64_t>(index));
-      vec->mutable_value()->data<T>()[row_index] +=
-          tmat.data<T>()[i * width + j];
-    }
-  }
-}
-
 template <typename T>
 void MatrixBitCodeFunctor<T>::Sum(const framework::Tensor& tmat,
                                   framework::Tensor* sum, T scale_sum) {

paddle/fluid/operators/math/matrix_bit_code.h

@@ -139,11 +139,11 @@ class SimpleCode : public Code {
 template <typename T>
 class CustomCode : public Code {
  public:
-  CustomCode(const framework::Tensor& ptable, const framework::Tensor& pcode,
-             const int64_t* ids, int index)
+  CustomCode(const framework::Tensor& path_table,
+             const framework::Tensor& path_code, const int64_t* ids, int index)
       : ids_(ids), index_(index) {
-    ptable_ = ptable.Slice(index, index + 1);
-    pcode_ = pcode.Slice(index, index + 1);
+    ptable_ = path_table.Slice(index, index + 1);
+    pcode_ = path_code.Slice(index, index + 1);
   }
   /**
    * Here the id of root shoud be 1 rather than 0, thus the encoding of class c
@@ -195,9 +195,9 @@ class SimpleCodeTable : public CodeTable {
 template <typename T>
 class CustomCodeTable : public CodeTable {
  public:
-  CustomCodeTable(const framework::Tensor& ptable,
-                  const framework::Tensor& pcode, const int64_t* ids)
-      : ptable_(ptable), pcode_(pcode), ids_(ids) {}
+  CustomCodeTable(const framework::Tensor& path_table,
+                  const framework::Tensor& path_code, const int64_t* ids)
+      : ptable_(path_table), pcode_(path_code), ids_(ids) {}
 
   std::unique_ptr<Code> get_code(int64_t code) const {
     std::unique_ptr<Code> coder(new CustomCode<T>(ptable_, pcode_, ids_, code));
@@ -223,11 +223,11 @@ class MatrixBitCodeFunctor {
         ids_(ids),
         code_table_(new SimpleCodeTable(num_classes, ids)) {}
 
-  MatrixBitCodeFunctor(const framework::Tensor& ptable,
-                       const framework::Tensor& pcode, const int64_t* ids)
-      : num_classes_(static_cast<size_t>(ptable.dims()[1])),
+  MatrixBitCodeFunctor(const framework::Tensor& path_table,
+                       const framework::Tensor& path_code, const int64_t* ids)
+      : num_classes_(static_cast<size_t>(path_table.dims()[1])),
         ids_(ids),
-        code_table_(new CustomCodeTable<int64_t>(ptable, pcode, ids)) {}
+        code_table_(new CustomCodeTable<int64_t>(path_table, path_code, ids)) {}
   /* For j < code_length
        tmat(i, j) += vec(0, index(i, j))
   */
@@ -238,11 +238,6 @@ class MatrixBitCodeFunctor {
   */
   void AddGrad(const framework::Tensor& tmat, framework::Tensor* vec);
 
-  /* For selected rows For j < code_length
-       vec(0, index(i, j)) += tmat(i, j)
-  */
-  void AddGrad(const framework::Tensor& tmat, framework::SelectedRows* vec);
-
   /* For j < code_length
     sum(i, 0) = \sum_j bit(i, j) * tmat(i, j)
   */

python/paddle/fluid/layers/nn.py

@@ -4931,6 +4931,9 @@ def hsigmoid(input,
         pass
 
     weights = None
+    remote_prefetch = False
+    if os.environ.get('PADDLE_ENABLE_REMOTE_PREFETCH'):
+        remote_prefetch = True
 
     if not is_custom:
         weights = helper.create_parameter(
@@ -4947,7 +4950,7 @@ def hsigmoid(input,
     inputs = {
         "X": input,
         "W": weights,
-        "PTable": path_table,
+        "PathTable": path_table,
         "PathCode": path_code,
         "Label": label
     }
@@ -4970,9 +4973,13 @@ def hsigmoid(input,
         type="hierarchical_sigmoid",
         inputs=inputs,
         outputs={"Out": out,
-                 "PreOut": pre_out},
-        attrs={"num_classes": num_classes,
-               "is_sparse": is_sparse})
+                 "PreOut": pre_out,
+                 "W_Out": weights},
+        attrs={
+            "num_classes": num_classes,
+            "is_sparse": is_sparse,
+            "remote_prefetch": remote_prefetch
+        })
     return out
 
 

python/paddle/fluid/tests/unittests/test_hsigmoid_op.py

@@ -185,7 +185,7 @@ class TestHSigmoidOpSparse(OpTest):
         self.inputs = {
             'X': x,
             'W': w,
-            'PTable': path_table,
+            'PathTable': path_table,
             'PathCode': path_code,
             'Label': label,
             'Bias': bias
@@ -287,7 +287,7 @@ class TestHSigmoidOpWithCostumTree(OpTest):
         self.inputs = {
             'X': x,
             'W': w,
-            'PTable': path_table,
+            'PathTable': path_table,
             'PathCode': path_code,
             'Label': label,
             'Bias': bias
@@ -324,7 +324,7 @@ class TestHSigmoidOpWithCostumTreeWithoutBias(OpTest):
         self.inputs = {
             'X': x,
             'W': w,
-            'PTable': path_table,
+            'PathTable': path_table,
             'PathCode': path_code,
             'Label': label,
         }