ARM int8 support

1. add fake_quant fake_dequant op
2. add quant_dequant fuse pass
3. fix bug for passes for arm
4. softmax axis problem

paddle/fluid/lite/api/cxx_api_bin.cc

@@ -86,6 +86,8 @@ USE_LITE_OP(depthwise_conv2d);
 USE_LITE_OP(pool2d);
 USE_LITE_OP(elementwise_add);
 USE_LITE_OP(softmax);
+USE_LITE_OP(fake_quantize_moving_average_abs_max);
+USE_LITE_OP(fake_dequantize_max_abs);
 
 USE_LITE_KERNEL(feed, kHost, kAny, kAny, def);
 USE_LITE_KERNEL(fetch, kHost, kAny, kAny, def);

paddle/fluid/lite/core/mir/CMakeLists.txt

@@ -9,6 +9,7 @@ cc_library(mir_passes
         SRCS fc_fuse_pass.cc
         conv_elementwise_add_relu_fuse_pass.cc
         conv_bn_fuse_pass.cc 
+        quant_dequant_fuse_pass.cc
         static_kernel_pick_pass.cc
         variable_place_inference_pass.cc
         type_target_transform_pass.cc

paddle/fluid/lite/core/mir/fusion/CMakeLists.txt

@@ -8,10 +8,15 @@ cc_library(fuse_conv_bn
         SRCS conv_bn_fuser.cc
         DEPS pattern_matcher_high_api)
 
+cc_library(fuse_quant_dequant
+        SRCS quant_dequant_op_fuser.cc
+        DEPS pattern_matcher_high_api)
+
 set(mir_fusers 
     fuse_fc 
     fuse_conv_elementwise_add_relu
     fuse_conv_bn
+    fuse_quant_dequant
     CACHE INTERNAL "fusers")
 
 if (LITE_WITH_LIGHT_WEIGHT_FRAMEWORK)

paddle/fluid/lite/core/mir/fusion/conv_elementwise_add_relu_fuser.cc

@@ -79,7 +79,7 @@ void ConvElementwiseAddReLUFuser::InsertNewNode(SSAGraph* graph,
 cpp::OpDesc ConvElementwiseAddReLUFuser::GenOpDesc(const key2nodes_t& matched) {
   auto* desc = matched.at("conv2d")->stmt()->op_info();
 
-  cpp::OpDesc op_desc;
+  cpp::OpDesc op_desc = *desc;
   op_desc.SetType(conv_type_);
   op_desc.SetInput("Input", {matched.at("input")->arg()->name});
   op_desc.SetInput("Filter", {matched.at("filter")->arg()->name});
@@ -92,7 +92,6 @@ cpp::OpDesc ConvElementwiseAddReLUFuser::GenOpDesc(const key2nodes_t& matched) {
                 "ResidualData") != input_arg_names.end()) {
     op_desc.SetInput("ResidualData", desc->Input("ResidualData"));
   }
-
   // Only consider strides, padding, groups, dilations, fuse_relu for now
   op_desc.SetAttr("strides", desc->GetAttr<std::vector<int>>("strides"));
   op_desc.SetAttr("paddings", desc->GetAttr<std::vector<int>>("paddings"));

paddle/fluid/lite/core/mir/fusion/quant_dequant_op_fuser.cc

+// 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/lite/core/mir/fusion/quant_dequant_op_fuser.h"
+#include <memory>
+#include <vector>
+
+namespace paddle {
+namespace lite {
+namespace mir {
+namespace fusion {
+
+void QuantDequantOpFuser::BuildPattern() {
+  const int kNumFields = 5;
+  const int kQuantizedWeightOffset = 0;
+  const int kQuantizedOpOffset = 1;
+  const int kQuantizedOpOutOffset = 2;
+  const int kDequantOpOffset = 3;
+  const int kDequantOpOutOffset = 4;
+
+  std::string weight_name = "";
+  if (op_type_ == "conv2d" || op_type_ == "depthwise_conv2d") {
+    weight_name = "Filter";
+  } else {
+    weight_name = "Y";
+  }
+  auto* quant_op_input = VarNode("quant_op_input")
+                             ->assert_is_op_input(quant_type_, "X")
+                             ->AsInput();
+  auto* quant_op_in_scale = VarNode("quant_op_in_scale")
+                                ->assert_is_op_input(quant_type_, "InScale")
+                                ->AsIntermediate();
+  auto* quant_op = OpNode("quant_op", quant_type_)
+                       ->assert_is_op(quant_type_)
+                       ->AsIntermediate();
+
+  auto* quant_op_out_scale =
+      VarNode("quant_op_out_scale")
+          ->assert_is_op_output(quant_type_, "OutScale")
+          ->assert_is_op_input("fake_dequantize_max_abs", "Scale")
+          ->AsIntermediate();
+
+  auto* quant_op_out = VarNode("quant_op_out")
+                           ->assert_is_op_output(quant_type_, "Out")
+                           ->assert_is_op_input(op_type_)
+                           ->AsIntermediate();
+  std::vector<PMNode*> nodes;
+  for (int i = 0; i < times_; i++) {
+    nodes.push_back(VarNode("quantized_op_weight" + std::to_string(i))
+                        ->assert_is_op_input(op_type_, weight_name)
+                        ->AsInput());
+
+    nodes.push_back(OpNode("quantized_op" + std::to_string(i), op_type_)
+                        ->assert_is_op(op_type_)
+                        ->AsIntermediate());
+
+    nodes.push_back(VarNode("quantized_op_out" + std::to_string(i))
+                        ->assert_is_op_output(op_type_)
+                        ->assert_is_op_input("fake_dequantize_max_abs", "X")
+                        ->AsIntermediate());
+
+    nodes.push_back(
+        OpNode("dequant_op" + std::to_string(i), "fake_dequantize_max_abs")
+            ->assert_is_op("fake_dequantize_max_abs")
+            ->AsIntermediate());
+    nodes.push_back(VarNode("dequant_op_out" + std::to_string(i))
+                        ->assert_is_op_output("fake_dequantize_max_abs", "Out")
+                        ->AsOutput());
+  }
+
+  quant_op->LinksFrom({quant_op_input, quant_op_in_scale});
+  quant_op_out->LinksFrom({quant_op});
+  quant_op_out_scale->LinksFrom({quant_op});
+  for (int i = 0; i < times_; i++) {
+    nodes[i * kNumFields + kQuantizedOpOffset]->LinksFrom(
+        {quant_op_out, nodes[i * kNumFields + kQuantizedWeightOffset]});
+    nodes[i * kNumFields + kQuantizedOpOutOffset]->LinksFrom(
+        {nodes[i * kNumFields + kQuantizedOpOffset]});
+    nodes[i * kNumFields + kDequantOpOffset]->LinksFrom(
+        {nodes[i * kNumFields + kQuantizedOpOutOffset], quant_op_out_scale});
+    nodes[i * kNumFields + kDequantOpOutOffset]->LinksFrom(
+        {nodes[i * kNumFields + kDequantOpOffset]});
+  }
+}
+
+void QuantDequantOpFuser::InsertNewNode(SSAGraph* graph,
+                                        const key2nodes_t& matched) {
+  const int kNumFields = 5;
+  const int kQuantizedWeightOffset = 0;
+  const int kQuantizedOpOffset = 1;
+  const int kQuantizedOpOutOffset = 2;
+  const int kDequantOpOffset = 3;
+  const int kDequantOpOutOffset = 4;
+
+  auto* quant_op_input = matched.at("quant_op_input");
+  auto* quant_op_in_scale = matched.at("quant_op_in_scale");
+  auto* quant_op = matched.at("quant_op");
+  auto* quant_op_out_scale = matched.at("quant_op_out_scale");
+  auto* quant_op_out = matched.at("quant_op_out");
+
+  std::vector<Node*> nodes;
+  for (int i = 0; i < times_; i++) {
+    nodes.push_back(matched.at("quantized_op_weight" + std::to_string(i)));
+    nodes.push_back(matched.at("quantized_op" + std::to_string(i)));
+    nodes.push_back(matched.at("quantized_op_out" + std::to_string(i)));
+    nodes.push_back(matched.at("dequant_op" + std::to_string(i)));
+    nodes.push_back(matched.at("dequant_op_out" + std::to_string(i)));
+  }
+  int bit_length = quant_op->stmt()->op_info()->GetAttr<int>("bit_length");
+  auto* scope = quant_op->stmt()->op->scope();
+  auto& valid_places = quant_op->stmt()->op->valid_places();
+  int range = ((1 << (bit_length - 1)) - 1);
+  auto input_scale_t = scope->FindVar(quant_op_in_scale->arg()->name)
+                           ->GetMutable<lite::Tensor>();
+  float input_scale = input_scale_t->data<float>()[0];
+
+  for (int i = 0; i < times_; i++) {
+    float max_range = nodes[i * kNumFields + kDequantOpOffset]
+                          ->stmt()
+                          ->op_info()
+                          ->GetAttr<float>("max_range");
+    float weight_scale = (range * range) / max_range;
+
+    cpp::OpDesc op_desc =
+        *nodes[i * kNumFields + kQuantizedOpOffset]->stmt()->op_info();
+    if (op_type_ == "conv2d" || op_type_ == "depthwise_conv2d") {
+      op_desc.SetInput("Input", {matched.at("quant_op_input")->arg()->name});
+      op_desc.SetOutput(
+          "Output", {nodes[i * kNumFields + kDequantOpOutOffset]->arg()->name});
+    } else if (op_type_ == "mul") {
+      op_desc.SetInput("X", {matched.at("quant_op_input")->arg()->name});
+      op_desc.SetOutput(
+          "Out", {nodes[i * kNumFields + kDequantOpOutOffset]->arg()->name});
+    }
+    op_desc.SetAttr("enable_int8", true);
+    op_desc.SetAttr("input_scale", input_scale);
+    auto quantized_weight_var_name =
+        nodes[i * kNumFields + kQuantizedWeightOffset]->arg()->name;
+    auto quantized_weight_t =
+        scope->FindVar(quantized_weight_var_name)->GetMutable<lite::Tensor>();
+    float* quantized_weight_data = quantized_weight_t->mutable_data<float>();
+    size_t weight_num = quantized_weight_t->data_size();
+    for (size_t i = 0; i < weight_num; i++) {
+      quantized_weight_data[i] *= (weight_scale / range);
+    }
+    auto quantized_op = LiteOpRegistry::Global().Create(op_type_);
+
+    quantized_op->Attach(op_desc, scope);
+    auto* new_op_node =
+        graph->GraphCreateInstructNode(quantized_op, valid_places);
+    IR_NODE_LINK_TO(quant_op_input, new_op_node);
+    IR_NODE_LINK_TO(nodes[i * kNumFields + kQuantizedWeightOffset],
+                    new_op_node);
+    IR_NODE_LINK_TO(new_op_node, nodes[i * kNumFields + kDequantOpOutOffset]);
+  }
+}
+
+cpp::OpDesc QuantDequantOpFuser::GenOpDesc(const key2nodes_t& matched) {
+  cpp::OpDesc op_desc;
+  return op_desc;
+}
+
+}  // namespace fusion
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/mir/fusion/quant_dequant_op_fuser.h

+// 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 <memory>
+#include <string>
+#include "paddle/fluid/lite/core/mir/pattern_matcher_high_api.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+namespace fusion {
+
+class QuantDequantOpFuser : public FuseBase {
+ public:
+  explicit QuantDequantOpFuser(const std::string& op_type,
+                               const std::string& quant_type, int times)
+      : op_type_(op_type), quant_type_(quant_type), times_(times) {}
+  void BuildPattern() override;
+  void InsertNewNode(SSAGraph* graph, const key2nodes_t& matched) override;
+
+ private:
+  cpp::OpDesc GenOpDesc(const key2nodes_t& matched) override;
+
+ private:
+  std::string op_type_{"conv2d"};
+  std::string quant_type_;
+  int times_;
+};
+
+}  // namespace fusion
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/mir/passes.h

@@ -22,6 +22,7 @@ namespace mir {}  // namespace mir
 }  // namespace paddle
 
 #ifndef LITE_WITH_LIGHT_WEIGHT_FRAMEWORK
+#endif
 USE_MIR_PASS(demo);
 USE_MIR_PASS(static_kernel_pick_pass);
 USE_MIR_PASS(variable_place_inference_pass);
@@ -29,9 +30,9 @@ USE_MIR_PASS(type_target_transform_pass);
 USE_MIR_PASS(generate_program_pass);
 USE_MIR_PASS(io_copy_kernel_pick_pass);
 USE_MIR_PASS(argument_type_display_pass);
-#endif
 USE_MIR_PASS(runtime_context_assign_pass);
 USE_MIR_PASS(lite_conv_bn_fuse_pass);
 USE_MIR_PASS(graph_visualze);
 USE_MIR_PASS(lite_fc_fuse_pass);
 USE_MIR_PASS(lite_conv_elementwise_add_act_fuse_pass);
+USE_MIR_PASS(lite_quant_dequant_fuse_pass);

paddle/fluid/lite/core/mir/pattern_matcher.cc

@@ -115,7 +115,6 @@ void PatternMatcher::operator()(SSAGraph *graph,
 bool PatternMatcher::MarkPMNodesInGraph(SSAGraph *graph) {
   VLOG(3) << "mark pmnodes in graph";
   if (graph->nodes().empty()) return false;
-
   for (auto &node : graph->mutable_nodes()) {
     for (const auto &pmnode : pattern_.nodes()) {
       if (pmnode->Tell(&node)) {
@@ -398,7 +397,7 @@ PMNode *PMNode::assert_is_op_output(const std::string &op_type) {
   asserts_.emplace_back([=](const Node *x) {
     for (auto *op : x->inlinks) {
       if (op && op->IsStmt()) {
-        auto *op_info = x->stmt()->op_info();
+        auto *op_info = op->stmt()->op_info();
         if (op_info->Type() == op_type) return true;
       }
     }

paddle/fluid/lite/core/mir/quant_dequant_fuse_pass.cc

+// 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/lite/core/mir/quant_dequant_fuse_pass.h"
+#include <memory>
+#include <vector>
+#include "paddle/fluid/lite/core/mir/fusion/quant_dequant_op_fuser.h"
+#include "paddle/fluid/lite/core/mir/pass_registry.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+
+void QuantDequantFusePass::Apply(const std::unique_ptr<SSAGraph>& graph) {
+  std::unordered_set<std::string> quant_types = {
+      "fake_quantize_range_abs_max", "fake_quantize_moving_average_abs_max"};
+  std::unordered_set<std::string> quantized_op_types = {"conv2d", "mul",
+                                                        "depthwise_conv2d"};
+  for (auto& quant_type : quant_types) {
+    for (auto& op_type : quantized_op_types) {
+      for (int i = 6; i >= 1; i--) {
+        fusion::QuantDequantOpFuser fuser(op_type, quant_type, i);
+        fuser(graph.get());
+      }
+    }
+  }
+}
+
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle
+
+REGISTER_MIR_PASS(lite_quant_dequant_fuse_pass,
+                  paddle::lite::mir::QuantDequantFusePass);

paddle/fluid/lite/core/mir/quant_dequant_fuse_pass.h

+// 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 <memory>
+#include <string>
+#include <unordered_set>
+#include "paddle/fluid/lite/core/mir/pass.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+
+class QuantDequantFusePass : public ProgramPass {
+ public:
+  void Apply(const std::unique_ptr<SSAGraph>& graph) override;
+};
+
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/optimizer.h

@@ -48,19 +48,19 @@ class Optimizer {
 
     if (passes.empty()) {
       RunPasses(std::vector<std::string>{{
+          "lite_quant_dequant_fuse_pass",             //
           "lite_conv_bn_fuse_pass",                   //
           "lite_conv_elementwise_add_act_fuse_pass",  //
           "lite_fc_fuse_pass",                        //
-#ifndef LITE_WITH_LIGHT_WEIGHT_FRAMEWORK
           "static_kernel_pick_pass",                  //
           "variable_place_inference_pass",            //
           "argument_type_display_pass",               //
           "type_target_transform_pass",               //
-          "argument_type_display_pass",     //
           "variable_place_inference_pass",            //
           "argument_type_display_pass",               //
           "io_copy_kernel_pick_pass",                 //
           "variable_place_inference_pass",            //
+#ifndef LITE_WITH_LIGHT_WEIGHT_FRAMEWORK
 #endif
           "runtime_context_assign_pass",  //
       }});

paddle/fluid/lite/core/target_wrapper.h

@@ -55,8 +55,8 @@ enum class DataLayoutType : int {
 #define DATALAYOUT(item__) paddle::lite::DataLayoutType::item__
 
 static const std::string& TargetToStr(TargetType target) {
-  static const std::string target2string[] = {"unk", "host", "x86", "cuda",
-                                              "any"};
+  static const std::string target2string[] = {"unk",  "host", "x86",
+                                              "cuda", "arm",  "any"};
   auto x = static_cast<int>(target);
   CHECK_LT(x, static_cast<int>(TARGET(NUM)));
   return target2string[x];

paddle/fluid/lite/core/type_system.h

@@ -165,8 +165,8 @@ class Type : public DataType {
 
 // -------------------------------- compatible check ---------------------------
 static bool TargetCompatibleTo(const Type& a, const Type& b) {
-  auto is_host = [](TargetType x) {
-    return x == TARGET(kHost) || x == TARGET(kX86);
+  auto is_host = [](TargetType x) -> bool {
+    return x == TARGET(kHost) || x == TARGET(kX86) || x == TARGET(kARM);
   };
   if (a.IsVoid() || b.IsVoid()) return true;
   if (a.IsTensor() || b.IsTensor()) {

paddle/fluid/lite/kernels/arm/conv_compute.cc

@@ -100,7 +100,7 @@ void ConvCompute::Run() {
 REGISTER_LITE_KERNEL(conv2d, kARM, kFloat, kNCHW,
                      paddle::lite::kernels::arm::ConvCompute, def)
     .BindInput("Input", {LiteType::GetTensorTy(TARGET(kARM))})
-    //  .BindInput("Bias", {LiteType::GetTensorTy(TARGET(kARM))})
+    .BindInput("Bias", {LiteType::GetTensorTy(TARGET(kARM))})
     .BindInput("Filter", {LiteType::GetTensorTy(TARGET(kARM))})
     .BindOutput("Output", {LiteType::GetTensorTy(TARGET(kARM))})
     .Finalize();
@@ -108,7 +108,7 @@ REGISTER_LITE_KERNEL(conv2d, kARM, kFloat, kNCHW,
 REGISTER_LITE_KERNEL(depthwise_conv2d, kARM, kFloat, kNCHW,
                      paddle::lite::kernels::arm::ConvCompute, def)
     .BindInput("Input", {LiteType::GetTensorTy(TARGET(kARM))})
-    //    .BindInput("Bias", {LiteType::GetTensorTy(TARGET(kARM))})
+    .BindInput("Bias", {LiteType::GetTensorTy(TARGET(kARM))})
     .BindInput("Filter", {LiteType::GetTensorTy(TARGET(kARM))})
     .BindOutput("Output", {LiteType::GetTensorTy(TARGET(kARM))})
     .Finalize();

paddle/fluid/lite/operators/CMakeLists.txt

@@ -21,6 +21,8 @@ cc_library(op_params_lite SRCS op_params.cc DEPS ${tensor_lite} any_lite framewo
 cc_library(dropout_op_lite SRCS dropout_op.cc DEPS ${op_DEPS})
 cc_library(concat_op_lite SRCS concat_op.cc DEPS ${op_DEPS})
 # cc_library(split_op_lite SRCS split_op.cc DEPS ${op_DEPS})
+cc_library(fake_quant SRCS fake_quantize_moving_avg_max_abs.cc DEPS ${op_DEPS})
+cc_library(fake_dequant SRCS fake_dequantize_max_abs.cc DEPS ${op_DEPS})
 
 set(ops_lite
         conv_op_lite
@@ -42,6 +44,8 @@ set(ops_lite
         dropout_op_lite
         concat_op_lite
         #split_op_lite
+        fake_quant
+        fake_dequant
         PARENT_SCOPE)
 
 lite_cc_test(test_fc_op_lite SRCS fc_op_test.cc 

paddle/fluid/lite/operators/fake_dequantize_max_abs.cc

+// 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/lite/operators/fake_dequantize_max_abs.h"
+#include "paddle/fluid/lite/core/op_registry.h"
+
+namespace paddle {
+namespace lite {
+namespace operators {}  // namespace operators
+}  // namespace lite
+}  // namespace paddle
+
+REGISTER_LITE_OP(fake_dequantize_max_abs,
+                 paddle::lite::operators::FakeDequantizeMaxAbsOpLite);

paddle/fluid/lite/operators/fake_dequantize_max_abs.h

+// 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 <string>
+#include <vector>
+#include "paddle/fluid/lite/core/compatible_tensor.h"
+#include "paddle/fluid/lite/core/kernel.h"
+#include "paddle/fluid/lite/core/op_lite.h"
+#include "paddle/fluid/lite/core/scope.h"
+#include "paddle/fluid/lite/operators/op_params.h"
+#include "paddle/fluid/lite/utils/all.h"
+
+namespace paddle {
+namespace lite {
+namespace operators {
+
+class FakeDequantizeMaxAbsOpLite : public OpLite {
+ public:
+  FakeDequantizeMaxAbsOpLite() {}
+
+  explicit FakeDequantizeMaxAbsOpLite(const std::string &type) : OpLite(type) {}
+
+  bool CheckShape() const override {}
+
+  bool InferShape() const override {}
+
+  bool AttachImpl(const cpp::OpDesc &op_desc, lite::Scope *scope) override {
+    auto x = op_desc.Input("X").front();
+    auto in_scale = op_desc.Input("Scale").front();
+
+    auto out = op_desc.Output("Out").front();
+
+    param_.x = scope->FindVar(x)->GetMutable<lite::Tensor>();
+    param_.in_scale = scope->FindVar(in_scale)->GetMutable<lite::Tensor>();
+
+    param_.out = scope->FindVar(out)->GetMutable<lite::Tensor>();
+    param_.max_range = op_desc.GetAttr<float>("max_range");
+    return true;
+  }
+
+  void AttachKernel(KernelBase *kernel) override { kernel->SetParam(param_); }
+
+  std::string DebugString() const override { return "fake_dequantize_max_abs"; }
+
+ private:
+  mutable FakeDequantizeMaxAbsParam param_;
+};
+
+}  // namespace operators
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/operators/fake_quantize_moving_avg_max_abs.cc

+// 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/lite/operators/fake_quantize_moving_avg_max_abs.h"
+#include "paddle/fluid/lite/core/op_registry.h"
+
+namespace paddle {
+namespace lite {
+namespace operators {}  // namespace operators
+}  // namespace lite
+}  // namespace paddle
+
+REGISTER_LITE_OP(fake_quantize_moving_average_abs_max,
+                 paddle::lite::operators::FakeQuantizeMovingAvgMaxAbsOpLite);

paddle/fluid/lite/operators/fake_quantize_moving_avg_max_abs.h

+// 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 <string>
+#include <vector>
+#include "paddle/fluid/lite/core/compatible_tensor.h"
+#include "paddle/fluid/lite/core/kernel.h"
+#include "paddle/fluid/lite/core/op_lite.h"
+#include "paddle/fluid/lite/core/scope.h"
+#include "paddle/fluid/lite/operators/op_params.h"
+#include "paddle/fluid/lite/utils/all.h"
+
+namespace paddle {
+namespace lite {
+namespace operators {
+
+class FakeQuantizeMovingAvgMaxAbsOpLite : public OpLite {
+ public:
+  FakeQuantizeMovingAvgMaxAbsOpLite() {}
+
+  explicit FakeQuantizeMovingAvgMaxAbsOpLite(const std::string &type)
+      : OpLite(type) {}
+
+  bool CheckShape() const override {}
+
+  bool InferShape() const override {}
+
+  bool AttachImpl(const cpp::OpDesc &op_desc, lite::Scope *scope) override {
+    auto x = op_desc.Input("X").front();
+    auto in_scale = op_desc.Input("InScale").front();
+
+    auto out = op_desc.Output("Out").front();
+    auto out_scale = op_desc.Output("OutScale").front();
+
+    param_.x = scope->FindVar(x)->GetMutable<lite::Tensor>();
+    param_.in_scale = scope->FindVar(in_scale)->GetMutable<lite::Tensor>();
+
+    param_.out = scope->FindVar(out)->GetMutable<lite::Tensor>();
+    param_.out_scale = scope->FindVar(out_scale)->GetMutable<lite::Tensor>();
+    param_.bit_length = op_desc.GetAttr<int>("bit_length");
+    return true;
+  }
+
+  void AttachKernel(KernelBase *kernel) override { kernel->SetParam(param_); }
+
+  std::string DebugString() const override {
+    return "fake_quantize_moving_avg_max_abs";
+  }
+
+ private:
+  mutable FakeQuantizeMovingAvgMaxAbsParam param_;
+};
+
+}  // namespace operators
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/operators/op_params.h

@@ -256,6 +256,28 @@ struct FillConstantParam {
   lite::Tensor* Out{};
 };
 
+//
+struct FakeQuantizeMovingAvgMaxAbsParam {
+  const lite::Tensor* x{};
+  const lite::Tensor* in_scale{};
+  const lite::Tensor* in_accum{};
+  const lite::Tensor* in_state{};
+  lite::Tensor* out{};
+  lite::Tensor* out_scale{};
+  lite::Tensor* out_state{};
+  lite::Tensor* out_accum{};
+  int bit_length;
+  bool is_test{true};
+  float moving_rate{0.9};
+};
+
+struct FakeDequantizeMaxAbsParam {
+  const lite::Tensor* x{};
+  const lite::Tensor* in_scale{};
+  lite::Tensor* out{};
+  float max_range;
+};
+
 /// ----------------------- sgd operators ----------------------
 struct SGDParam {
   int dtype{framework::proto::VarType::FP32};

paddle/fluid/lite/operators/softmax_op.cc

@@ -39,7 +39,12 @@ bool SoftmaxOp::AttachImpl(const cpp::OpDesc &opdesc, lite::Scope *scope) {
       &scope->FindVar(opdesc.Input("X").front())->Get<lite::Tensor>());
   param_.output =
       scope->FindVar(opdesc.Output("Out").front())->GetMutable<lite::Tensor>();
+
+  if (opdesc.HasAttr("axis")) {
     param_.axis = opdesc.GetAttr<int>("axis");
+  } else {
+    param_.axis = -1;
+  }
   CHECK(param_.x);
   CHECK(param_.output);
   return true;