[CodeStyle][Typos][A-[21-27],A-[29-30]] Fix typos (allctor,almostly,alog,Aread,Allways,alway,ane,adn,expaned,annotaions,annotaion,architecure,architecuture,aer)

_typos.toml

@@ -19,6 +19,7 @@ pash = 'pash'
 astroid = 'astroid'
 eles = 'eles'
 builer = 'builer'
+anc = 'anc'
 
 # These words need to be fixed
 ontext = 'ontext'
@@ -71,7 +72,6 @@ bootom = 'bootom'
 Vetical = 'Vetical'
 mantain = 'mantain'
 patition = 'patition'
-almostly = 'almostly'
 Dowloading = 'Dowloading'
 Prepar = 'Prepar'
 precent = 'precent'
@@ -122,7 +122,6 @@ decalared = 'decalared'
 coalesc = 'coalesc'
 graident = 'graident'
 infered = 'infered'
-Allways = 'Allways'
 substitue = 'substitue'
 Ouput = 'Ouput'
 witk = 'witk'
@@ -132,7 +131,6 @@ staticly = 'staticly'
 emited = 'emited'
 repalce = 'repalce'
 GARD = 'GARD'
-annotaions = 'annotaions'
 gloabl = 'gloabl'
 devide = 'devide'
 zerp = 'zerp'
@@ -225,8 +223,6 @@ Rto = 'Rto'
 tunning = 'tunning'
 kerenl = 'kerenl'
 Temperarily = 'Temperarily'
-alway = 'alway'
-ane = 'ane'
 cliping = 'cliping'
 DEIVCE = 'DEIVCE'
 neeed = 'neeed'
@@ -541,7 +537,6 @@ instrinsics = 'instrinsics'
 outputing = 'outputing'
 hadler = 'hadler'
 qucik = 'qucik'
-alog = 'alog'
 exsit = 'exsit'
 deciamls = 'deciamls'
 uncorrectly = 'uncorrectly'
@@ -644,7 +639,6 @@ insid = 'insid'
 coodinate = 'coodinate'
 usefull = 'usefull'
 sqaure = 'sqaure'
-adn = 'adn'
 intialize = 'intialize'
 Taget = 'Taget'
 parm = 'parm'
@@ -730,7 +724,6 @@ compitable = 'compitable'
 comple = 'comple'
 dealed = 'dealed'
 ser = 'ser'
-anc = 'anc'
 contraints = 'contraints'
 propogated = 'propogated'
 beacuse = 'beacuse'
@@ -795,7 +788,6 @@ imformation = 'imformation'
 kernerl = 'kernerl'
 Boardcast = 'Boardcast'
 Greate = 'Greate'
-Alread = 'Alread'
 unkown = 'unkown'
 recevied = 'recevied'
 Normlized = 'Normlized'
@@ -804,7 +796,6 @@ orginal = 'orginal'
 Stati = 'Stati'
 Destory = 'Destory'
 seperately = 'seperately'
-alloctor = 'alloctor'
 fullfill = 'fullfill'
 Substitude = 'Substitude'
 producted = 'producted'

paddle/cinn/operator_fusion/graph_transformer/search_algorithm.h

@@ -140,8 +140,8 @@ struct SearchAlgorithm<ReverseTopoNodePairPattern,
 template <typename Kind, typename GraphMatcher, typename GraphOperation>
 void GraphTransformer(PatternGraph* graph) {
   VLOG(4) << "Start GraphTransformer...";
-  auto alog = SearchAlgorithm<Kind, GraphMatcher, GraphOperation>(graph);
-  alog();
+  auto algo = SearchAlgorithm<Kind, GraphMatcher, GraphOperation>(graph);
+  algo();
 }
 
 }  // namespace cinn::fusion

paddle/fluid/operators/fused/fused_attention_op.cc

@@ -389,8 +389,8 @@ class FusedAttentionOpMaker : public framework::OpProtoAndCheckerMaker {
                   "The qkv_w shape is (h, 3h), do transpose to it.")
         .SetDefault(false);
     AddAttr<bool>("pre_layer_norm",
-                  "if true, the attention op uses pre_layer_norm architecure, "
-                  "else, uses post_layer_norm architecuture. "
+                  "if true, the attention op uses pre_layer_norm architecture, "
+                  "else, uses post_layer_norm architecture. "
                   "[default false].")
         .SetDefault(false);
     AddAttr<float>("epsilon",

paddle/fluid/platform/init.cc

@@ -294,7 +294,7 @@ void SignalHandle(const char *data, int size) {
     // NOTE1: The glog FailureSignalHandler dumped messages
     //   are deal with line by line
     auto signal_msg_dumper_ptr = SignalMessageDumper::Instance().Get();
-    // NOTE2: we only deal with the time info ane signal info,
+    // NOTE2: we only deal with the time info and signal info,
     //   the stack trace will generated by paddle self
     if (StartsWith(data, "*** Aborted at")) {
       *signal_msg_dumper_ptr << "\n  [TimeInfo: " << std::string(data, size - 1)

paddle/fluid/pybind/pybind.cc

@@ -2476,7 +2476,7 @@ All parameter, weight, gradient are variables in Paddle.
 
   m.def("_is_program_version_supported", IsProgramVersionSupported);
 #if defined(PADDLE_WITH_CUDA)
-  m.def("alloctor_dump", [](const phi::GPUPlace &place) {
+  m.def("allocator_dump", [](const phi::GPUPlace &place) {
     auto allocator = std::dynamic_pointer_cast<
         paddle::memory::allocation::AutoGrowthBestFitAllocator>(
         paddle::memory::allocation::AllocatorFacade::Instance()

paddle/phi/infermeta/spmd_rules/cross_entropy_with_softmax.cc

@@ -189,7 +189,7 @@ SpmdInfo CrossEntropyWithSoftmaxInferSpmdBase(const DistMetaTensor& x,
   // todo if softmax_normalize axis is sharded, notify downstream phi api to
   // select c_softmax_with_entropy_kernel.
 
-  // according to the phi api implementation, the softmax_out tensor will alway
+  // according to the phi api implementation, the softmax_out tensor will always
   // be generated not matter the value of use_softmax.
   return {{x_dist_attr_dst, label_dist_attr_dst},
           {softmax_out_dist_attr_dst, loss_dist_attr_dst}};
@@ -363,7 +363,7 @@ SpmdInfo CrossEntropyWithSoftmaxInferSpmdReverse(
           << str_join(x_dims_mapping) << "]\nLabel dims_mapping: ["
           << str_join(label_dims_mapping) << "]\n\n";
 
-  // according to the phi api implementation, the softmax_out tensor will alway
+  // according to the phi api implementation, the softmax_out tensor will always
   // be generated not matter the value of use_softmax.
   return {{x_dist_attr, label_dist_attr},
           {s_out_dist_attr_dst, loss_dist_attr_dst}};

python/paddle/distributed/passes/auto_parallel_fp16.py

@@ -124,7 +124,7 @@ def _keep_fp32_input(op, in_name):
 
 # TODO check if bf16 and fp16 still share the same logic
 def _keep_fp32_output(op, out_name):
-    # TODO(lizhiyu02): Support 'custom_white_list' adn 'custom_black_list' in amp_options
+    # TODO(lizhiyu02): Support 'custom_white_list' and 'custom_black_list' in amp_options
     if not op.amp_options.enable:
         return True
     op_type = op.type

python/paddle/incubate/asp/supported_layer_list.py

@@ -72,7 +72,7 @@ def _default_pruning(
     # In sparse training, layer weight matrices is viewed sparse matrix A, so
     # the math formula should be 'Act(WX + b)'. However, default formula in PaddlePaddle
     #  is 'Act(XW + b)'. For enabling SPMMA, weights and inputs should be transposed
-    # for computing, Act( (W^T X^T)^T + b). Therefore, we have to prune alog k dimension
+    # for computing, Act( (W^T X^T)^T + b). Therefore, we have to prune along k dimension
     # of W^T, which is m dimension of W. Moreover, all mask generating functions in
     # asp/utils is row-major pruning. That is the reason we have to transpose weight
     # matrices before invoking create_mask. Then we transpose the result mask to make

python/paddle/nn/utils/transform_parameters.py

@@ -33,7 +33,7 @@
     from paddle._typing import ShapeLike
 
 
-# input==output, inplace strategy of reshape has no cost almostly
+# input==output, inplace strategy of reshape has no cost almost
 def _inplace_reshape_dygraph(x: Tensor, shape: ShapeLike) -> None:
     x_shape = _create_tensor(dtype='int64')
     if in_dygraph_mode():

python/paddle/optimizer/optimizer.py

@@ -1157,7 +1157,7 @@ def _create_optimization_pass(
         # _create_accumulators method if it needs to create accumulators
         # for parameters and extend _finish_update method to add custom ops.
 
-        # Allways called under program_guard use global block as loss block
+        # Always called under program_guard use global block as loss block
         # But if current block is in control flow, append optimize op in the
         # grad block of current block
 

python/paddle/static/quantization/post_training_quantization.py

@@ -354,7 +354,7 @@ def __init__(
         self._quantized_weight_var_name = set()
         self._quantized_act_var_name = set()
         self._weight_op_pairs = {}
-        # The vars for alog = KL or hist
+        # The vars for algo = KL or hist
         self._sampling_act_abs_min_max = {}
         self._sampling_act_histogram = {}
         self._sampling_data = {}

python/paddle/tensor/tensor.prototype.pyi

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 # The `Tensor` template `tensor.prototype.pyi` for `tools/gen_tensor_stub.py` to generate the stub file `tensor.pyi`.
-# Add docstring, attributes, methods and alias with type annotaions for `Tensor` in `tensor.prototype.pyi`
+# Add docstring, attributes, methods and alias with type annotations for `Tensor` in `tensor.prototype.pyi`
 # if not conveniently coding in original place (like c++ source file).
 
 # Import common typings for generated methods

test/cpp/fluid/platform/errors_test.cc

@@ -92,7 +92,7 @@ TEST(Errors, NotFound) { CHECK_ALL_PADDLE_EXCEPTION_MACRO(NotFound); }
 
 TEST(Errors, OutOfRange) { CHECK_ALL_PADDLE_EXCEPTION_MACRO(OutOfRange); }
 
-TEST(Errors, AlreadExists) { CHECK_ALL_PADDLE_EXCEPTION_MACRO(AlreadyExists); }
+TEST(Errors, AlreadyExists) { CHECK_ALL_PADDLE_EXCEPTION_MACRO(AlreadyExists); }
 
 TEST(Errors, ResourceExhausted) {
   CHECK_ALL_PADDLE_EXCEPTION_MACRO(ResourceExhausted);

test/cpp/phi/kernels/test_gpu_timer.cu

@@ -86,19 +86,19 @@ TEST(GpuTimer, Sum) {
 #endif
 
   using Functor = std::function<void(float *, float *, size_t)>;
-  Functor alog0 = Algo<4, 256, 1024>;
+  Functor algo0 = Algo<4, 256, 1024>;
   Functor algo1 = Algo<1, 256, 1024>;
-  Functor alog2 = Algo<1, 256, 8>;
+  Functor algo2 = Algo<1, 256, 8>;
 
-  std::vector<Functor> algos = {alog0, algo1, alog2};
+  std::vector<Functor> algos = {algo0, algo1, algo2};
 
   for (int j = 0; j < algos.size(); ++j) {
     auto algo = algos[j];
     phi::GpuTimer timer;
     timer.Start(0);
     algo(d_in1, d_in2, N);
     timer.Stop(0);
-    VLOG(3) << "alog: " << j << " cost: " << timer.ElapsedTime() << "ms";
+    VLOG(3) << "algo: " << j << " cost: " << timer.ElapsedTime() << "ms";
   }
 
 #ifdef __HIPCC__