Fix some typos (manully, scpoe_feed_vec_, etc.) (#61740)

co63oc · web-flow · commit 42b91db15f20 · 2024-02-19T15:55:46.000+08:00
diff --git a/paddle/fluid/framework/custom_operator.cc b/paddle/fluid/framework/custom_operator.cc
@@ -545,7 +545,7 @@ static void RunInferShapeFunc(
               "Custom operator only supports `paddle::Vec(...)` inputs and "
               "cannot support `paddle::Vec(...)` output without setting "
               "InplaceMap. If you have to use `paddle::Vec(...)` output, "
-              "please indicate it by setting InplaceMap manully."));
+              "please indicate it by setting InplaceMap manually."));
       // make sure ctx has valid inplace optional outputs
       if (ctx->HasOutputs(out_name)) {
         auto in_name = inplace_reverse_map.at(out_name);
@@ -796,7 +796,7 @@ static void RunInferDtypeFunc(
               "Custom operator only supports `paddle::Vec(...)` inputs and "
               "cannot support `paddle::Vec(...)` output without setting "
               "InplaceMap. If you have to use `paddle::Vec(...)` output, "
-              "please indicate it by setting InplaceMap manully."));
+              "please indicate it by setting InplaceMap manually."));
       auto in_name = inplace_reverse_map.at(out_name);
       // make sure ctx has valid inplace optional outputs
       if (ctx->HasOutput(out_name)) {
diff --git a/paddle/fluid/framework/data_feed.cc b/paddle/fluid/framework/data_feed.cc
@@ -440,7 +440,7 @@ int InMemoryDataFeed<T>::Next() {
     }
     VLOG(3) << "enable heter next: " << offset_index_
             << " batch_offsets: " << batch_offsets_.size()
-            << " baych_size: " << this->batch_size_;
+            << " batch_size: " << this->batch_size_;
   }
   return this->batch_size_;
 #else
@@ -2514,10 +2514,10 @@ bool SlotRecordInMemoryDataFeed::ParseOneInstance(const std::string& line,
 void SlotRecordInMemoryDataFeed::AssignFeedVar(const Scope& scope) {
   CheckInit();
 #if defined(PADDLE_WITH_CUDA) && defined(PADDLE_WITH_HETERPS)
-  if (scpoe_feed_vec_.count(&scope) > 0) {
+  if (scope_feed_vec_.count(&scope) > 0) {
     return;
   }
-  auto& feed_vec = scpoe_feed_vec_[&scope];
+  auto& feed_vec = scope_feed_vec_[&scope];
   feed_vec.resize(used_slots_info_.size());
   for (int i = 0; i < use_slot_size_; ++i) {
     feed_vec[i] =
@@ -2869,8 +2869,8 @@ void SlotRecordInMemoryDataFeed::BuildSlotBatchGPU(const int ins_num,
   auto* dev_ctx = static_cast<phi::GPUContext*>(
       platform::DeviceContextPool::Instance().Get(this->place_));
   for (int j = 0; j < use_slot_size_; ++j) {
-    if (scpoe_feed_vec_.size() > 0) {
-      if (scpoe_feed_vec_.begin()->second[j] == nullptr) {
+    if (scope_feed_vec_.size() > 0) {
+      if (scope_feed_vec_.begin()->second[j] == nullptr) {
         h_tensor_ptrs[j] = nullptr;
         continue;
       }
@@ -2952,8 +2952,8 @@ void SlotRecordInMemoryDataFeed::PackToScope(MiniBatchGpuPack* pack,
 
   auto* feed_vec = &feed_vec_;
   if (scope) {
-    CHECK(scpoe_feed_vec_.count(scope) > 0) << "scope not found.";
-    feed_vec = &scpoe_feed_vec_[scope];
+    CHECK(scope_feed_vec_.count(scope) > 0) << "scope not found.";
+    feed_vec = &scope_feed_vec_[scope];
   }
 
   CHECK(feed_vec != nullptr) << "feed_vec nullptr.";
diff --git a/paddle/fluid/framework/data_feed.h b/paddle/fluid/framework/data_feed.h
@@ -876,7 +876,7 @@ struct BufState {
     len = tmp_len;
     central_word = -1;
     step = -1;
-    GetNextCentrolWord();
+    GetNextCentralWord();
   }
 
   int GetNextStep() {
@@ -895,7 +895,7 @@ struct BufState {
             << " row_num: " << row_num;
   }
 
-  int GetNextCentrolWord() {
+  int GetNextCentralWord() {
     if (++central_word >= walk_len) {
       return 0;
     }
@@ -923,7 +923,7 @@ struct BufState {
     len = tmp_len;
     central_word = -1;
     step = -1;
-    GetNextCentrolWord();
+    GetNextCentralWord();
     return tmp_len != 0;
   }
 };
@@ -1412,7 +1412,7 @@ class DataFeed {
   bool train_mode_;
 };
 
-// PrivateQueueDataFeed is the base virtual class for ohther DataFeeds.
+// PrivateQueueDataFeed is the base virtual class for other DataFeeds.
 // It use a read-thread to read file and parse data to a private-queue
 // (thread level), and get data from this queue when trainer call Next().
 template <typename T>
@@ -2000,7 +2000,7 @@ class SlotRecordInMemoryDataFeed : public InMemoryDataFeed<SlotRecord> {
   std::mutex pack_mutex_;
 
   // async infershape
-  std::map<const Scope*, std::vector<phi::DenseTensor*>> scpoe_feed_vec_;
+  std::map<const Scope*, std::vector<phi::DenseTensor*>> scope_feed_vec_;
 #endif
 };
 
diff --git a/paddle/fluid/framework/data_set.cc b/paddle/fluid/framework/data_set.cc
@@ -1288,7 +1288,7 @@ int MultiSlotDataset::ReceiveFromClient(int msg_type,
     index = global_index_++;
   }
   index = index % channel_num_;
-  VLOG(3) << "ramdom index=" << index;
+  VLOG(3) << "random index=" << index;
   multi_output_channel_[index]->Write(std::move(data));
 
   data.clear();
diff --git a/paddle/fluid/framework/data_set.h b/paddle/fluid/framework/data_set.h
@@ -154,7 +154,7 @@ class Dataset {
   virtual void ClearLocalTables() = 0;
   // create preload readers
   virtual void CreatePreLoadReaders() = 0;
-  // destroy preload readers after prelaod done
+  // destroy preload readers after preload done
   virtual void DestroyPreLoadReaders() = 0;
   // set preload thread num
   virtual void SetPreLoadThreadNum(int thread_num) = 0;
diff --git a/python/paddle/distributed/auto_parallel/static/operators/dist_dropout.py b/python/paddle/distributed/auto_parallel/static/operators/dist_dropout.py
@@ -130,7 +130,7 @@ def forward(ctx, *args, **kwargs):
                 and src_op.attr("seed")
             ):
                 _logger.info(
-                    f"Auto Parallel Random Control Skipped Since manul seed is set by user: {src_op}"
+                    f"Auto Parallel Random Control Skipped Since manual seed is set by user: {src_op}"
                 )
             elif rank_id not in op_dist_attr.process_mesh.process_ids:
                 pass
@@ -161,7 +161,7 @@ def forward(ctx, *args, **kwargs):
                     pre_op._set_attr("force_cpu", True)
                 else:
                     _logger.info(
-                        f"Auto Parallel Random Control Skipped Since manul seed is set by user: {src_op}"
+                        f"Auto Parallel Random Control Skipped Since manual seed is set by user: {src_op}"
                     )
             else:
                 # determinate rng
diff --git a/python/paddle/distributed/auto_parallel/static/operators/dist_eltwise.py b/python/paddle/distributed/auto_parallel/static/operators/dist_eltwise.py
@@ -61,7 +61,7 @@ def update_dims_mapping(dist_op):
         output_arg_name = op_desc.output_arg_names()[0]
         num_inputs = len(input_arg_names)
 
-        # TODO (zhangyichen) replace dist tensor spece by dist tensor in future.
+        # TODO (zhangyichen) replace dist tensor specs by dist tensor in future.
         input_specs = []
         for i in range(num_inputs):
             input_specs.append(
diff --git a/python/paddle/distributed/auto_parallel/static/operators/dist_embedding.py b/python/paddle/distributed/auto_parallel/static/operators/dist_embedding.py
@@ -473,7 +473,7 @@ def forward(ctx, *args, **kwargs):
                 ctx, op_dist_attr.process_mesh, rank_id
             )
 
-        # A generalized method to calculate embedding offset using cartisian product
+        # A generalized method to calculate embedding offset using cartesian product
         relative_idx = _get_idx_in_axis(
             process_mesh_group,
             process_mesh_shape,
@@ -632,7 +632,7 @@ def backward(ctx, *args, **kwargs):
         process_mesh_shape = dist_attr.process_mesh.shape
         process_mesh_group = dist_attr.process_mesh.process_ids
 
-        # A generalized method to calculate embedding offset using cartisian product
+        # A generalized method to calculate embedding offset using cartesian product
         relative_idx = _get_idx_in_axis(
             process_mesh_group,
             process_mesh_shape,
diff --git a/python/paddle/distributed/auto_parallel/static/operators/dist_fused_dropout_add.py b/python/paddle/distributed/auto_parallel/static/operators/dist_fused_dropout_add.py
@@ -87,7 +87,7 @@ def forward(ctx, *args, **kwargs):
                 and src_op.attr("seed")
             ):
                 _logger.info(
-                    f"Auto Parallel Random Control Skipped Since manul seed is set by user: {src_op}"
+                    f"Auto Parallel Random Control Skipped Since manual seed is set by user: {src_op}"
                 )
             elif rank_id not in op_dist_attr.process_mesh.process_ids:
                 pass
@@ -118,7 +118,7 @@ def forward(ctx, *args, **kwargs):
                     pre_op._set_attr("force_cpu", True)
                 else:
                     _logger.info(
-                        f"Auto Parallel Random Control Skipped Since manul seed is set by user: {src_op}"
+                        f"Auto Parallel Random Control Skipped Since manual seed is set by user: {src_op}"
                     )
             else:
                 # determinate rng
diff --git a/python/paddle/distributed/fleet/dataset/dataset.py b/python/paddle/distributed/fleet/dataset/dataset.py
@@ -49,7 +49,7 @@ def init(
         download_cmd="cat",
     ):
         """
-        should be called only once in user's python scripts to initialize setings of dataset instance.
+        should be called only once in user's python scripts to initialize settings of dataset instance.
         Normally, it is called by InMemoryDataset or QueueDataset.
 
         Args:
@@ -280,7 +280,7 @@ def _check_use_var_with_data_generator(
         self, var_list, data_generator_class, test_file
     ):
         """
-         Var consistency insepection of use_var_list and data_generator data.
+         Var consistency inspection of use_var_list and data_generator data.
 
         Examples:
             .. code-block:: python
@@ -382,7 +382,7 @@ def _init_distributed_settings(self, **kwargs):
         """
         :api_attr: Static Graph
 
-        should be called only once in user's python scripts to initialize distributed-related setings of dataset instance
+        should be called only once in user's python scripts to initialize distributed-related settings of dataset instance
         Args:
             kwargs: Keyword arguments. Currently, we support following keys in **kwargs:
 
@@ -444,7 +444,7 @@ def update_settings(self, **kwargs):
         """
         :api_attr: Static Graph
 
-        should be called in user's python scripts to update setings of dataset instance.
+        should be called in user's python scripts to update settings of dataset instance.
 
         Args:
             kwargs: Keyword arguments. Currently, we support following keys in **kwargs,
@@ -525,7 +525,7 @@ def init(self, **kwargs):
         """
         :api_attr: Static Graph
 
-        should be called only once in user's python scripts to initialize setings of dataset instance
+        should be called only once in user's python scripts to initialize settings of dataset instance
 
         Args:
             kwargs: Keyword arguments. Currently, we support following keys in **kwargs:
@@ -1316,7 +1316,7 @@ def init(self, **kwargs):
         """
         :api_attr: Static Graph
 
-        should be called only once in user's python scripts to initialize setings of dataset instance
+        should be called only once in user's python scripts to initialize settings of dataset instance
 
         """
         super().init(**kwargs)
@@ -1356,7 +1356,7 @@ def __init__(self):
 
     def init(self, **kwargs):
         """
-        should be called only once in user's python scripts to initialize setings of dataset instance
+        should be called only once in user's python scripts to initialize settings of dataset instance
         """
         super().init(**kwargs)
 
@@ -1382,7 +1382,7 @@ def __init__(self):
 
     def init(self, **kwargs):
         """
-        should be called only once in user's python scripts to initialize setings of dataset instance
+        should be called only once in user's python scripts to initialize settings of dataset instance
         """
         super().init(**kwargs)
 
diff --git a/python/paddle/distributed/fleet/elastic/manager.py b/python/paddle/distributed/fleet/elastic/manager.py
@@ -59,7 +59,7 @@ def __init__(self, args):
         self.procs = []
 
     def _terminate_procs(self):
-        # try to terminate process by group, this happend in multiprocess senario in user process
+        # try to terminate process by group, this happened in multiprocess scenario in user process
         if os.name != 'nt':
             for p in self.procs:
                 if p.proc.poll() is None:
@@ -79,7 +79,7 @@ def _terminate_procs(self):
         for step in range(0, 50):
             alive = False
             for p in self.procs:
-                if p.proc.poll() is None:  # not termniate
+                if p.proc.poll() is None:  # not terminate
                     os.kill(p.proc.pid, signal.SIGKILL)
                     alive = True
 
@@ -186,7 +186,7 @@ def __init__(self, args, etcd_client):
             self.elastic_level = ElasticLevel.ELASTIC
             logger.info('start job with ElasticLevel.ELASTIC')
 
-        # compatible with kuberntes service discovery
+        # compatible with kubernetes service discovery
         if (
             not server
             and os.getenv('PADDLE_ELASTIC_ETCD_SERVICE_HOST')
@@ -457,7 +457,7 @@ def _update_endpoint(self, endpoints, hosts):
             f'{endpoints}|{hosts}'.encode('latin-1'),
         )
 
-    def _update_fault_tolrance(self):
+    def _update_fault_tolerance(self):
         rank = int(os.getenv('PADDLE_TRAINER_ID', -1))
         logger.debug(
             f"self.curr_host={self.curr_host}, self.dist_endpoints={self.dist_endpoints}"
@@ -559,12 +559,12 @@ def _update_elastic_scale_in(self):
     def _update_hosts(self):
         assert len(self.hosts) != 0, 'hosts empty'
         if self.elastic_level == ElasticLevel.FAULT_TOLERANCE:
-            self._update_fault_tolrance()
+            self._update_fault_tolerance()
         else:
             # elastic
             if len(self.hosts) == self.np:
                 logger.info(f"elastic startup, hosts={self.hosts}")
-                self._update_fault_tolrance()
+                self._update_fault_tolerance()
 
             elif len(self.hosts) > self.np:
                 # scale out
diff --git a/python/paddle/distributed/fleet/layers/mpu/mp_layers.py b/python/paddle/distributed/fleet/layers/mpu/mp_layers.py
@@ -342,7 +342,7 @@ class ColumnParallelLinear(paddle.nn.Layer):
         weight_attr(ParamAttr|None): The attribute for the learnable weight of this layer. The default value is None
             and the weight will be initialized to zero. For detailed information, please refer to paddle.ParamAttr.
         has_bias(bool): whether to add bias.
-        gather_output(bool): whether to do allgahter for the output of each rank.
+        gather_output(bool): whether to do allgather for the output of each rank.
         fuse_matmul_bias(bool): whether to fuse matmul and bias.
         mp_group(Group): The tensor parallel group.
         name(str, optional): Normally there is no need for user to set this parameter.
@@ -549,7 +549,7 @@ class RowParallelLinear(paddle.nn.Layer):
         weight_attr(ParamAttr|None): The attribute for the learnable weight of this layer. The default value is None
             and the weight will be initialized to zero. For detailed information, please refer to paddle.ParamAttr.
         has_bias(bool): whether to add bias.
-        input_is_parallel(bool): whether the input has alreadly been splitted across the mp group.
+        input_is_parallel(bool): whether the input has already been splitted across the mp group.
         fuse_matmul_bias(bool): whether to fuse matmul and bias.
         mp_group(Group): The tensor parallel group.
         name(str, optional): Normally there is no need for user to set this parameter.
@@ -758,7 +758,7 @@ class ParallelCrossEntropy(paddle.nn.Layer):
             >>> # doctest: +SKIP('No img to demonstrate')
             >>> from paddle.distributed.fleet.layers.mpu import ParallelCrossEntropy
             >>> loss_func = ParallelCrossEntropy
-            >>> loss = loss_func(img, lable)
+            >>> loss = loss_func(img, label)
 
     """
 
diff --git a/python/paddle/distributed/fleet/layers/mpu/mp_ops.py b/python/paddle/distributed/fleet/layers/mpu/mp_ops.py
diff --git a/python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py b/python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
diff --git a/python/paddle/distributed/fleet/meta_optimizers/sharding/utils.py b/python/paddle/distributed/fleet/meta_optimizers/sharding/utils.py

Original file line number	Diff line number	Diff line change
`@@ -1288,7 +1288,7 @@ int MultiSlotDataset::ReceiveFromClient(int msg_type,`
`1288`	`1288`	`index = global_index_++;`
`1289`	`1289`	`}`
`1290`	`1290`	`index = index % channel_num_;`
`1291`		`- VLOG(3) << "ramdom index=" << index;`
	`1291`	`+ VLOG(3) << "random index=" << index;`
`1292`	`1292`	`multi_output_channel_[index]->Write(std::move(data));`
`1293`	`1293`
`1294`	`1294`	`data.clear();`