fix typo words (#22653)

paddle/fluid/framework/data_set.h

@@ -126,7 +126,7 @@ class Dataset {
   virtual void DestroyPreLoadReaders() = 0;
   // set preload thread num
   virtual void SetPreLoadThreadNum(int thread_num) = 0;
-  // seperate train thread and dataset thread
+  // separate train thread and dataset thread
   virtual void DynamicAdjustChannelNum(int channel_num) = 0;
   virtual void DynamicAdjustReadersNum(int thread_num) = 0;
   // set fleet send sleep seconds

paddle/fluid/framework/details/build_strategy.h

@@ -133,10 +133,10 @@ struct BuildStrategy {
   // The picture is here:
   // https://github.com/PaddlePaddle/Paddle/pull/17263#discussion_r285411396
   bool use_hierarchical_allreduce_{false};
-  // Nccl ranks in a node when use hierarchical allreduce, it's setted to gpu
+  // Nccl ranks in a node when use hierarchical allreduce, it's set to gpu
   // cards' number in most cases.
   size_t hierarchical_allreduce_inter_nranks_{0};
-  // Nccl ranks bewteen nodes when use hierarchical allreduce, it's setted to
+  // Nccl ranks bewteen nodes when use hierarchical allreduce, it's set to
   // nodes number.
   size_t hierarchical_allreduce_exter_nranks_{0};
 

paddle/fluid/framework/ir/conv_elementwise_add2_act_fuse_pass.cc

@@ -33,7 +33,7 @@ namespace ir {
   GET_IR_NODE(act_op);                 \
   GET_IR_NODE(act_out);
 
-// Inherient the basic infomation from `base_desc`, and modify some fields.
+// Inherient the basic information from `base_desc`, and modify some fields.
 framework::proto::OpDesc PrepareOpDesc(
     const framework::proto::OpDesc& base_desc, const std::string& bias,
     const std::string& bias1, const std::string& activation,

paddle/fluid/framework/ir/conv_elementwise_add_act_fuse_pass.cc

@@ -31,7 +31,7 @@ namespace ir {
   GET_IR_NODE(act_op);               \
   GET_IR_NODE(act_out);
 
-// Inherient the basic infomation from `base_desc`, and modify some fields.
+// Inherient the basic information from `base_desc`, and modify some fields.
 framework::proto::OpDesc PrepareOpDesc(
     const framework::proto::OpDesc& base_desc, const std::string& bias,
     const std::string& activation, const std::string& output) {

paddle/fluid/framework/ir/fuse_optimizer_ops_pass/fuse_optimizer_op_pass.cc

@@ -382,7 +382,7 @@ const VarDesc *FuseOptimizerOpPass::GetVarDescFromVarsInfo(
     const std::string &var_name) const {
   auto grad_iter = vars_info.find(var_name);
   PADDLE_ENFORCE_EQ(grad_iter != vars_info.end(), true,
-                    "The gradient varibale %s is not found.", var_name);
+                    "The gradient variable %s is not found.", var_name);
   PADDLE_ENFORCE_EQ(!grad_iter->second.empty(), true,
                     "The gradient var node %s is not found.", var_name);
   PADDLE_ENFORCE_NOT_NULL(grad_iter->second.front()->Var(),

paddle/fluid/framework/ir/graph_pattern_detector.cc

@@ -131,7 +131,7 @@ bool GraphPatternDetector::MarkPDNodesInGraph(const ir::Graph &graph) {
 }
 
 // The intermediate Nodes can only link to the nodes inside the pattern, or this
-// subgraph will be droped.
+// subgraph will be dropped.
 void GraphPatternDetector::ValidateByNodeRole(
     std::vector<GraphPatternDetector::subgraph_t> *subgraphs) {
   std::vector<GraphPatternDetector::subgraph_t> result;

paddle/fluid/framework/ir/multi_batch_merge_pass.cc

@@ -179,7 +179,7 @@ void BatchMergePass::ApplyImpl(ir::Graph* graph) const {
         ir::Node* var = nullptr;
         auto updated_var = UpdateGradVarDesc(in_node->Var(), i, grad_names,
                                              bn_vars_need_rename);
-        // should be initialized by startup, how to initilize tensor in the
+        // should be initialized by startup, how to initialize tensor in the
         // scope?
         if (node->Name() == "batch_norm" &&
             bn_vars_need_rename.find(in_node->Name()) !=

paddle/fluid/framework/ir/multi_devices_graph_pass/multi_devices_graph_pass.cc

@@ -1041,7 +1041,7 @@ void DistSSAGraphBuilder::InsertPostprocessOps(ir::Graph *result) const {
     // There are 4 conditions:
     // 1. GPU && Reduce: Reduce gradient then broadcast gradient to other GPUS.
     // Need to broadcast received parameters to other GPU.
-    // 2. GPU && AllReduce: AllReduce all graident to each GPU. Need to
+    // 2. GPU && AllReduce: AllReduce all gradient to each GPU. Need to
     // broadcast received parameters to other GPU.
     // 3. CPU && AllReduce: AllReduce all gradient to each thread. Need to
     // broadcast received parameters to other scope.

paddle/fluid/framework/op_desc.cc

@@ -80,7 +80,7 @@ class CompileTimeInferShapeContext : public InferShapeContext {
     PADDLE_ENFORCE_EQ(
         in_var_names.size(), out_var_names.size(),
         platform::errors::PreconditionNotMet(
-            "Op [%s]:  Input var number shoule be equal with output var number",
+            "Op [%s]:  Input var number should be equal with output var number",
             op_.Type()));
 
     for (size_t i = 0; i < in_var_names.size(); ++i) {
@@ -663,7 +663,7 @@ void OpDesc::Flush() {
 
 void OpDesc::CheckAttrs() {
   PADDLE_ENFORCE(!Type().empty(),
-                 "CheckAttr() can not be called before type is setted.");
+                 "CheckAttr() can not be called before type is set.");
   auto *checker = OpInfoMap::Instance().Get(Type()).Checker();
   if (checker == nullptr) {
     // checker is not configured. That operator could be generated by Paddle,
@@ -706,7 +706,7 @@ void OpDesc::InferShape(const BlockDesc &block) const {
 void OpDesc::InferVarType(BlockDesc *block) const {
   // There are a few places that var type can be set.
   // When VarDesc is created, default set to LOD_TENSOR.
-  // When output variable is created, default is defaut set to LOD_TENSOR.
+  // When output variable is created, default is default set to LOD_TENSOR.
   // We limit here to be the only place that operator defines its customized
   // var type inference. Hence, we don't do any "default" setting here.
   auto &info = OpInfoMap::Instance().Get(this->Type());

paddle/fluid/framework/operator.cc

@@ -654,7 +654,7 @@ class RuntimeInferShapeContext : public InferShapeContext {
     PADDLE_ENFORCE_EQ(
         in_var_list.size(), out_var_list.size(),
         platform::errors::PreconditionNotMet(
-            "Op [%s]: Input var size should be equal with ouput var size",
+            "Op [%s]: Input var size should be equal with output var size",
             op_.Type()));
 
     auto& out_var_names = op_.Outputs(out);

paddle/fluid/framework/operator.h

@@ -53,8 +53,8 @@ constexpr char kEmptyVarName[] = "@EMPTY@";
 constexpr char kTempVarName[] = "@TEMP@";
 
 /// If a variable's name has a certain suffix, it means that the
-/// variable is the gradient of another varibale.
-/// e.g. Variable "x@GRAD" is the gradient of varibale "x".
+/// variable is the gradient of another variable.
+/// e.g. Variable "x@GRAD" is the gradient of variable "x".
 constexpr char kGradVarSuffix[] = "@GRAD";
 
 constexpr size_t kGradVarSuffixSize = 5U;

paddle/fluid/framework/operator_test.cc

@@ -340,7 +340,7 @@ class IndicateLoDTensorDataTypeTestProtoMaker : public OpProtoAndCheckerMaker {
  public:
   void Make() {
     AddInput("LoDTensor", "Input of Tensor type Variable.");
-    AddComment("This Op is only for IndicateVarDataType inferface test.");
+    AddComment("This Op is only for IndicateVarDataType interface test.");
   }
 };
 
@@ -362,7 +362,7 @@ class IndicateSelectedRowsDataTypeTestProtoMaker
  public:
   void Make() {
     AddInput("SelectedRows", "Input of SelectedRows type Variable.");
-    AddComment("This Op is only for IndicateVarDataType inferface test.");
+    AddComment("This Op is only for IndicateVarDataType interface test.");
   }
 };
 
@@ -382,7 +382,7 @@ class IndicateOtherDataTypeTestProtoMaker : public OpProtoAndCheckerMaker {
  public:
   void Make() {
     AddInput("Other", "Input of Other type Variable");
-    AddComment("This Op is only for IndicateVarDataType inferface test.");
+    AddComment("This Op is only for IndicateVarDataType interface test.");
   }
 };
 
@@ -572,7 +572,7 @@ class GetSetLoDLevelTestMaker : public OpProtoAndCheckerMaker {
   void Make() {
     AddInput("X", "(LoDTensor) Input Variable.");
     AddOutput("Out", "(LoDTensor) Output Variable.");
-    AddComment("This Op is only for Get/SetLoDLevel inferface test.");
+    AddComment("This Op is only for Get/SetLoDLevel interface test.");
   }
 };
 

paddle/fluid/inference/analysis/ir_passes/subgraph_util.cc

@@ -112,7 +112,7 @@ void RenameAndGetOutputs(
     std::unordered_map<std::string, std::string> *output_name_map,
     const std::unordered_map<std::string, framework::ir::Node *> &graph_var_map,
     bool trt_and_not_int8) {
-  //// In the normal case, the paddle-trt exists bug when runing the googlenet.
+  //// In the normal case, the paddle-trt exists bug when running the googlenet.
   // When there are more than two convolutions of 1 * 1 with the same input, the
   // paddle-tensorrt will do the merging optimization, which fuse those conv
   // into one conv, and then trigger bug. So,  We should use strategy to avoid

paddle/fluid/inference/analysis/ir_passes/tensorrt_subgraph_pass.cc

@@ -223,7 +223,7 @@ void TensorRtSubgraphPass::CreateTensorRTOp(
   auto use_static_engine = Get<bool>("use_static_engine");
   // TODO(NHZlX)
   // There are models with the same structure but the different parameters,
-  // when runing in the 'use_serialize' mode, there is a bug.
+  // when running in the 'use_serialize' mode, there is a bug.
   auto engine_key = GenerateEngineKey(input_names_with_id, output_names_with_id,
                                       std::to_string(0));
   auto predictor_id = Get<int>("predictor_id");

paddle/fluid/inference/io.cc

@@ -137,7 +137,7 @@ std::unique_ptr<framework::ProgramDesc> Load(framework::Executor* executor,
                  "model version %ld is not supported.",
                  main_program->Version());
 
-  // model_from_memory is false in seperate parameters.
+  // model_from_memory is false in separate parameters.
   LoadPersistables(executor, scope, *main_program, dirname, "",
                    false /* model_from_memory */);
   return main_program;

paddle/fluid/inference/tensorrt/convert/ut_helper.h

@@ -101,7 +101,7 @@ class TRTConvertValidation {
     DeclVar(name, dim_vec);
   }
 
-  // Declare a parameter varaible in the scope.
+  // Declare a parameter variable in the scope.
   void DeclParamVar(const std::string& name, const nvinfer1::Dims& dims) {
     DeclVar(name, dims, true);
   }

paddle/fluid/inference/tensorrt/engine.cc

@@ -104,10 +104,9 @@ void TensorRTEngine::FreezeNetwork() {
 
       for (auto &t : all_t) {
         if (!quant_dynamic_range_.count(t)) {
-          VLOG(3)
-              << "We are in trt int8 mode(not calibration), scale not setted"
-              << " for tensor " << t->getName()
-              << ", this might be ok when trt does not need this range";
+          VLOG(3) << "We are in trt int8 mode(not calibration), scale not set"
+                  << " for tensor " << t->getName()
+                  << ", this might be ok when trt does not need this range";
         }
       }
       std::unordered_set<std::string> all_out_t_name;

paddle/fluid/operators/array_to_lod_tensor_op.cc

@@ -172,7 +172,7 @@ class ArrayToLoDTensorOpProtoMaker : public framework::OpProtoAndCheckerMaker {
              "(std::vector<LodTensor>) A vector of tensors that is going to "
              "be casted to a big LoDTensor.");
     AddInput("RankTable",
-             "(LoDRankTable) RankTable provides the coarse lod infomation to "
+             "(LoDRankTable) RankTable provides the coarse lod information to "
              "build the output LoDTensor. See "
              "'paddle/framework/lod_rank_table.h' for more details.");
     AddOutput("Out", "(LoDTensor) The LoDTensor formed by input tensor array.");

paddle/fluid/operators/average_accumulates_op.cc

@@ -132,7 +132,7 @@ class AverageAccumulatesOpMaker : public framework::OpProtoAndCheckerMaker {
         "(Tensor<int64_t>), The accumulating times of previous window with "
         "shape [1].");
     AddInput("in_num_updates",
-             "(Tensor<int64_t>), The total number of batches used by trainning "
+             "(Tensor<int64_t>), The total number of batches used by training "
              "before this batch with shape [1].");
 
     AddOutput("out_sum_1",
@@ -155,10 +155,9 @@ class AverageAccumulatesOpMaker : public framework::OpProtoAndCheckerMaker {
         "out_old_num_accumulates",
         "(Tensor<int64_t>) The accumulating times of previous window with "
         "shape [1].");
-    AddOutput(
-        "out_num_updates",
-        "(Tensor<int64_t>), The total number of batches used by trainning "
-        "before this batch with shape [1].");
+    AddOutput("out_num_updates",
+              "(Tensor<int64_t>), The total number of batches used by training "
+              "before this batch with shape [1].");
 
     AddAttr<float>("average_window",
                    "(float, default 0) "

paddle/fluid/operators/bilinear_tensor_product_op.h

@@ -49,7 +49,7 @@ class BilinearTensorProductKernel : public framework::OpKernel<T> {
     auto& place = *ctx.template device_context<DeviceContext>().eigen_device();
     auto& dev_ctx = ctx.template device_context<DeviceContext>();
 
-    // Create the intermediate variable to caculate the result of
+    // Create the intermediate variable to calculate the result of
     // Input(X) multiplied by Input(Weight_i), the formula is:
     // left_mul = X Weight_i.
     Tensor left_mul;

paddle/fluid/operators/conv_transpose_op.cc

@@ -267,7 +267,7 @@ void Conv2DTransposeOpMaker::Make() {
                "workspace is a section of GPU memory which will be "
                "allocated/freed each time the operator runs, larger "
                "workspace size can increase performance but also requires "
-               "better hardward. This size should be carefully setted.")
+               "better hardward. This size should be carefully set.")
       .SetDefault(platform::GetDefaultConvWorkspaceSizeLimitMB());
   AddComment(R"DOC(
 Convolution2D Transpose Operator.
@@ -368,7 +368,7 @@ void Conv3DTransposeOpMaker::Make() {
                "workspace is a section of GPU memory which will be "
                "allocated/freed each time the operator runs, larger "
                "workspace size can increase performance but also requires "
-               "better hardward. This size should be carefully setted.")
+               "better hardward. This size should be carefully set.")
       .SetDefault(platform::GetDefaultConvWorkspaceSizeLimitMB());
   AddComment(R"DOC(
 Convolution3D Transpose Operator.

paddle/fluid/operators/crop_op.cc

@@ -36,7 +36,7 @@ class CropOp : public framework::OperatorWithKernel {
       auto shape = ctx->Attrs().Get<std::vector<int>>("shape");
       PADDLE_ENFORCE_EQ(
           int64_t(shape.size()), x_dim.size(),
-          "Shape size should be equal to dimention size of input tensor.");
+          "Shape size should be equal to dimension size of input tensor.");
       std::vector<int64_t> tensor_shape(shape.size());
       for (size_t i = 0; i < shape.size(); ++i) {
         tensor_shape[i] = static_cast<int64_t>(shape[i]);

paddle/fluid/operators/crop_tensor_op.cc

@@ -82,7 +82,7 @@ class CropTensorOp : public framework::OperatorWithKernel {
     }
     PADDLE_ENFORCE_EQ(int64_t(shape.size()), x_dim.size(),
                       "Attr(shape)'size of Op(crop_tensor) should be equal to "
-                      "dimention size of input tensor.");
+                      "dimension size of input tensor.");
     std::vector<int64_t> out_shape(shape.size(), -1);
     for (size_t i = 0; i < shape.size(); ++i) {
       if (shape[i] > 0) {

paddle/fluid/operators/crop_tensor_op.h

@@ -157,7 +157,7 @@ void CropTensorFunction(const framework::ExecutionContext& context) {
 
   // get shape from Input(ShapeTensor) of Input(Shape)
   std::vector<int> shape = GetShape(context);
-  // out_dims setted by arrt(shape)
+  // out_dims set by arrt(shape)
   if (shape.size() == 0) {
     for (int i = 0; i < out_dims.size(); ++i) {
       shape.push_back(out_dims[i]);

paddle/fluid/operators/cross_entropy_op.cc

@@ -203,7 +203,7 @@ class CrossEntropyOpMaker : public framework::OpProtoAndCheckerMaker {
               "represents the cross entropy loss.");
     AddAttr<bool>("soft_label",
                   "(bool, default false), a flag indicating whether to "
-                  "interpretate the given labels as soft labels.")
+                  "interpretant the given labels as soft labels.")
         .SetDefault(false);
     AddAttr<int>("ignore_index",
                  "(int, default -100), Specifies a target value that is"