[CodeStyle][Typos][D-[37-44]] Fix typos (dito,devide,documention,doens,doen,dobule,doubel,dowloading,downsteram)

_typos.toml

@@ -36,15 +36,6 @@ unpacket = "unpacket"
 # These words need to be fixed
 Creater = 'Creater'
 creater = 'creater'
-dito = 'dito'
-devide = 'devide'
-documention = 'documention'
-doens = 'doens'
-doen = 'doen'
-doubel = 'doubel'
-dobule = 'dobule'
-Dowloading = 'Dowloading'
-downsteram = 'downsteram'
 fetchs = 'fetchs'
 Flattend = 'Flattend'
 flattend = 'flattend'

paddle/cinn/operator_fusion/graph_transformer/matcher.h

@@ -285,9 +285,9 @@ struct LeafReshapeConnectionMatcher {
 struct NotAllElementWiseDownstreamMatcher {
   bool operator()(const PatternGraph& graph, const PatternNodePtr& node) {
     size_t count = 0;
-    for (const auto& downsteram : node->downstream()) {
-      if (StmtPatternGraphMatcher<TrivialPattern>()(graph, downsteram)) {
-        auto ops = std::get<TrivialPattern>(downsteram->stmt_pattern()).ops();
+    for (const auto& downstream : node->downstream()) {
+      if (StmtPatternGraphMatcher<TrivialPattern>()(graph, downstream)) {
+        auto ops = std::get<TrivialPattern>(downstream->stmt_pattern()).ops();
         bool is_elementwise =
             std::all_of(ops.begin(), ops.end(), [](pir::Operation* op) {
               return GetOpPatternKind(op) == hlir::framework::kElementWise;

paddle/fluid/inference/tensorrt/op_teller.cc

@@ -2383,7 +2383,7 @@ struct SimpleOpTypeSetTeller : public Teller {
 
     // conv3d_transpose
     if (op_type == "conv3d_transpose") {
-      // trt doen't support output_padding when < 8406
+      // trt doesn't support output_padding when < 8406
       // output_padding is usually set when stride > 1
 #if !IS_TRT_VERSION_GE(8400)
       if (desc.HasAttr("output_padding")) {

paddle/fluid/operators/controlflow/control_flow_op_helper.h

@@ -121,7 +121,7 @@ static void AssignZeroToParentScope(
       PADDLE_ENFORCE_EQ(input_tensors.size(),
                         outside_tensors->size(),
                         common::errors::InvalidArgument(
-                            "DenseTensorArray outside_var %s doen't have same "
+                            "DenseTensorArray outside_var %s doesn't have same "
                             "size as input_var %s.",
                             outside_grad_name,
                             input_name));

paddle/fluid/operators/elementwise/elementwise_mul_op.cc

@@ -139,7 +139,7 @@ class ElementwiseMulCompositeDoubleGradOpMaker
                       -1,
                       common::errors::InvalidArgument(
                           "We only support axis = -1 in composite "
-                          "add_doubel_grad but we got: ",
+                          "add_double_grad but we got: ",
                           axis));
 
     // get output

paddle/fluid/operators/elementwise/elementwise_sub_op.cc

@@ -125,7 +125,7 @@ class ElementwiseSubCompositeDoubleGradOpMaker
                       -1,
                       common::errors::InvalidArgument(
                           "We only support axis = -1 in composite "
-                          "subtract_doubel_grad but we got: ",
+                          "subtract_double_grad but we got: ",
                           axis));
 
     paddle::Tensor* grad_out_grad = this->GetOutputPtr(&grad_out_grad_t);

paddle/fluid/pir/dialect/op_generator/op_gen.py

@@ -342,7 +342,7 @@ class {TEST_API} {op_name} : public pir::Op<{op_name}{interfaces}{traits}> {{
     'int': 'pir::Int32Attribute',
     'int64_t': 'pir::Int64Attribute',
     'float': 'pir::FloatAttribute',
-    'dobule': 'pir::DoubleAttribute',
+    'double': 'pir::DoubleAttribute',
     'bool': 'pir::BoolAttribute',
 }
 

paddle/phi/kernels/funcs/values_vectors_functor.h

@@ -467,7 +467,6 @@ struct MatrixEighFunctor<GPUContext, T> {
                                   "When has_vectors is true,"
                                   "the eigenvectors needs to be calculated,"
                                   "so the eigenvectors must be provided."));
-      //   input_trans = dito.Transpose(input_trans);
       input_trans = phi::TransposeLast2Dim<T>(dev_ctx, input_trans);
       eigen_vectors->ShareDataWith(input_trans);
     }

paddle/scripts/paddle_build.sh

@@ -3674,7 +3674,7 @@ function distribute_test() {
     parallel_fa_unit
     echo "End FA tests"
 
-    echo "Dowloading ...."
+    echo "Downloading ...."
     cd ${work_dir}
     wget https://paddlenlp.bj.bcebos.com/wheels/PaddleNLP_stable_paddle.tar.gz --no-proxy
     tar -zvxf PaddleNLP_stable_paddle.tar.gz

paddle/utils/string/printf.h

@@ -54,7 +54,7 @@
 //                  weekday, month, day, hour, min);
 //
 // 2. High-performance -- most printed strings are not too long and
-//    doens't need dynamic memory allocation.  Many StringPrintf
+//    doesn't need dynamic memory allocation.  Many StringPrintf
 //    implementations doesn't enforce type-safe, but are
 //    high-performance, including
 //

python/paddle/distributed/auto_parallel/static/tuner/to_distributed_api_patterns.py

@@ -563,7 +563,7 @@ def apply(
         key_states = paddle.transpose(key_states, [0, 2, 1, 3])
         value_states = paddle.transpose(value_states, [0, 2, 1, 3])
 
-        # matmul and devide by sqrt(head_dim)
+        # matmul and divide by sqrt(head_dim)
         attn_weights = paddle.matmul(
             query_states / math.sqrt(head_dim),
             key_states.transpose([0, 1, 3, 2]),

python/paddle/nn/clip.py

@@ -1407,7 +1407,7 @@ def set_gradient_clip(clip, param_list=None, program=None):
         "We recommend a new strategy: set 'grad_clip' "
         "when initializing the 'optimizer'. "
         "This method can reduce the mistakes, please "
-        "refer to documention of 'optimizer'."
+        "refer to documentation of 'optimizer'."
     )
 
     if not isinstance(clip, ClipGradBase):

test/auto_parallel/hybrid_strategy/to_distributed_api_for_llama.py

@@ -117,7 +117,7 @@ def scaled_dot_product_attention(
     key_states = paddle.transpose(key_states, [0, 2, 1, 3])
     value_states = paddle.transpose(value_states, [0, 2, 1, 3])
 
-    # matmul and devide by sqrt(head_dim)
+    # matmul and divide by sqrt(head_dim)
     attn_weights = paddle.matmul(
         query_states / math.sqrt(head_dim), key_states.transpose([0, 1, 3, 2])
     )

test/ir/pir/cinn/llama_test_model.py

@@ -187,7 +187,7 @@ def scaled_dot_product_attention(
     key_states = paddle.transpose(key_states, [0, 2, 1, 3])
     value_states = paddle.transpose(value_states, [0, 2, 1, 3])
 
-    # matmul and devide by sqrt(head_dim)
+    # matmul and divide by sqrt(head_dim)
     attn_weights = paddle.matmul(
         query_states / math.sqrt(head_dim), key_states.transpose([0, 1, 3, 2])
     )

test/legacy_test/test_cond.py

@@ -471,7 +471,7 @@ def test_extremely_simple_net_with_op_in_condition(self):
                 main_program, fetch_list=[out, b, a.grad_name, b.grad_name]
             )
         # Note: fill_constant has loss of precision, you have to assertEqual
-        # with values doens't lose precision in float-point number.
+        # with values doesn't lose precision in float-point number.
         self.assertEqual(ret[0][0], ret[1][0])
         self.assertEqual(ret[2][0], 0.0)
         self.assertEqual(ret[3][0], 1.0)