add fake_quantization op conversion in onnx

oneflow/core/job_rewriter/quantization_aware_training.cpp

@@ -124,7 +124,7 @@ std::string OpTypeName4OpNode(const OpNode* node) {
 
 using OpConfMap = HashMap<std::string, OperatorConf>;
 
-template<DataType VALUE_TYPE = DataType::kFloat>
+template<DataType data_type = DataType::kFloat>
 OperatorConf Get1DZeroVariableOpConf(std::string name, const int64_t scope_symbol_id,
                                      const int64_t length, OpConfMap* inserted_ops) {
   OperatorConf variable_op_conf{};
@@ -133,7 +133,7 @@ OperatorConf Get1DZeroVariableOpConf(std::string name, const int64_t scope_symbo
   VariableOpConf* variable_conf = variable_op_conf.mutable_variable_conf();
   variable_conf->set_out("out");
   *variable_conf->mutable_shape()->mutable_dim()->Add() = length;
-  variable_conf->set_data_type(VALUE_TYPE);
+  variable_conf->set_data_type(data_type);
   variable_conf->mutable_initializer()->mutable_constant_conf()->set_value(0);
   (*inserted_ops)[name] = variable_op_conf;
   return variable_op_conf;
@@ -255,22 +255,19 @@ user_op::UserOpConfWrapper MovingMinMaxObserver(const std::string& name, const s
       Get1DZeroVariableOpConf(moving_max_name, scope_symbol_id, 1, inserted_ops);
   const auto moving_min_var =
       Get1DZeroVariableOpConf(moving_min_name, scope_symbol_id, 1, inserted_ops);
+  std::string train_step_value = train_step_lbn;
+  if (!GlobalJobDesc().IsTrain()) {
+    const std::string train_step_name = name + TRAIN_STEP_SUFFIX;
+    const auto train_step_var = Get1DZeroVariableOpConf<DataType::kInt64>(
+        train_step_name, scope_symbol_id, 1, inserted_ops);
+    train_step_value =
+        GenLogicalBlobName(train_step_var.name(), train_step_var.variable_conf().out());
+  }
   const auto op_wrapper =
       user_op::UserOpConfWrapperBuilder(name)
           .Op("moving_average_min_max_observer")
           .Input("in", input)
-          .Input("current_train_step",
-                 [&] {
-                   const std::string train_step_name = name + TRAIN_STEP_SUFFIX;
-                   const auto train_step_var = Get1DZeroVariableOpConf<DataType::kInt64>(
-                       train_step_name, scope_symbol_id, 1, inserted_ops);
-                   if (GlobalJobDesc().IsTrain()) {
-                     return train_step_lbn;
-                   } else {
-                     return GenLogicalBlobName(train_step_var.name(),
-                                               train_step_var.variable_conf().out());
-                   }
-                 }())
+          .Input("current_train_step", train_step_value)
           .Input("moving_max",
                  GenLogicalBlobName(moving_max_var.name(), moving_max_var.variable_conf().out()))
           .Input("moving_min",