[Fix] Fix some typos (apache#11503)

Fix some typos in src/.

Co-authored-by: driazati <[email protected]>

Author: Yulv-git

NEWS.md

@@ -1262,7 +1262,7 @@ The community also continues to bring high quality improvements to the existing
 * Fixed div by zero core dump. Fixed rounding intrinsics on int crash #5026
 * Test case modified for int type #5012
 * Bug Fix for ARM CPUs. Lower strict assumption. #5063
-* Triage the testcases to fit the the new namespaces #5071
+* Triage the testcases to fit the new namespaces #5071
 * Add colors to `compute_at` edges and thread/block indices. #5111
 * Temporary fix to the stack overflow issue in autotvm task extraction #5019
 * Fix compilation of If-Elses #5040
@@ -2223,7 +2223,7 @@ Rust language support in TVM includes two parts. 1. The frontend wraps the curre
 ### Bug Fixes
 * [RELAY] Fix `get_int_tuple`. (#2691)
 * [ARITH] Select support for integer set analysis. (#2687)
-* [Relay] Fix error in ANF (too agressively inline atomic expression and create free variable). (#2665)
+* [Relay] Fix error in ANF (too aggressively inline atomic expression and create free variable). (#2665)
 * [Hybrid Script] Fix name conflict and attached scope problem. (#2649)
 * [Relay] Fix ANF for reference and pattern matching. (#2637)
 * [Relay] Fix fusion bug when call symbol that is not an operator. (#2630)

apps/ios_rpc/tvmrpc/TVMRuntime.mm

@@ -31,7 +31,7 @@
 #include <../../../src/runtime/file_utils.h>
 
 #if defined(USE_CUSTOM_DSO_LOADER) && USE_CUSTOM_DSO_LOADER == 1
-// internal TVM header to achive Library class
+// internal TVM header to achieve Library class
 #include <../../../src/runtime/library_module.h>
 #include <custom_dlfcn.h>
 #endif

apps/topi_recipe/README.md

@@ -26,7 +26,7 @@ and optimizing tvm generated kernels. The goal:
 
 ## Guidelines
 - Use numpy-style naming convention for known ops
-- Seperate operator declaration from schedule when possible.
+- Separate operator declaration from schedule when possible.
   - This can be inconvenient but enables more general scheduling across ops.
   - We can always recover the tensors from its outputs by traversing the tree.
 - Deliberately assert the requirements

cmake/modules/HexagonSDK.cmake

@@ -44,7 +44,7 @@ function(_check_all_paths_exist _paths _output_variable)
     if(_out_path)
       list(APPEND _out_paths "${_out_path}")
     else()
-      set_parent(${_ouput_variable} "${_path}-NOTFOUND")
+      set_parent(${_output_variable} "${_path}-NOTFOUND")
       return()
     endif()
   endforeach()

docs/arch/pass_infra.rst

@@ -51,7 +51,7 @@ scheme through `Sequential`_ and `Block`_, respectively. With such constructs,
 these modern frameworks are able to conveniently add modules/layers to their
 containers and build up neural networks easily.
 
-The design of the Relay pass infra is largely inspired by the the hierarchical
+The design of the Relay pass infra is largely inspired by the hierarchical
 pass manager used in LLVM and the block-style containers used in the popular
 deep learning frameworks. The major goals of the pass infra include:
 

docs/arch/security.rst

@@ -28,7 +28,7 @@ We strongly encourage folks to report such problems to our private security mail
 
 Please note that the security mailing list should only be used for reporting undisclosed security vulnerabilities and managing the process of fixing such vulnerabilities. We cannot accept regular bug reports or other queries at this address. All mail sent to this address that does not relate to an undisclosed security problem in our source code will be ignored.
 Questions about: if a vulnerability applies to your particular application obtaining further information on a published vulnerability availability of patches
-and/or new releases should be addressed to to the user discuss forum.
+and/or new releases should be addressed to the user Discuss forum.
 
 The private security mailing address is: `[email protected] <[email protected]>`_.
 Feel free to consult the `Apache Security guide <https://www.apache.org/security/>`_.

docs/how_to/deploy/index.rst

@@ -70,7 +70,7 @@ After you get the TVM runtime library, you can link the compiled library
 
 A model (optimized or not by TVM) can be cross compiled by TVM for
 different architectures such as ``aarch64`` on a ``x64_64`` host. Once the model
-is cross compiled it is neccessary to have a runtime compatible with the target
+is cross compiled it is necessary to have a runtime compatible with the target
 architecture to be able to run the cross compiled model.
 
 

docs/index.rst

@@ -18,7 +18,7 @@
 Apache TVM Documentation
 ========================
 
-Welcome to the the documentation for Apache TVM, a deep learning compiler that
+Welcome to the documentation for Apache TVM, a deep learning compiler that
 enables access to high-performance machine learning anywhere for everyone.
 TVM's diverse community of hardware vendors, compiler engineers and ML
 researchers work together to build a unified, programmable software stack, that

docs/reference/langref/hybrid_script.rst

@@ -110,7 +110,7 @@ In HalideIR, loops have in total 4 types: ``serial``, ``unrolled``, ``parallel``
 
 Here we use ``range`` aka ``serial``, ``unroll``, ``parallel``, and ``vectorize``,
 these **4** keywords to annotate the corresponding types of for loops.
-The the usage is roughly the same as Python standard ``range``.
+The usage is roughly the same as Python standard ``range``.
 
 Besides all the loop types supported in Halide, ``const_range`` is supported for some specific conditions.
 Sometimes, ``tvm.container.Array`` is desired to pass as an argument, but in TVM-HalideIR, there is no

gallery/how_to/optimize_operators/opt_gemm.py

@@ -346,7 +346,7 @@
 ###################################################################################################
 # Parallel
 # --------
-# Futhermore, we can also utilize multi-core processors to do the thread-level parallelization.
+# Furthermore, we can also utilize multi-core processors to do the thread-level parallelization.
 
 s = te.create_schedule(C.op)
 

include/tvm/ir/type.h

@@ -425,7 +425,7 @@ class FuncTypeNode : public TypeNode {
   Array<TypeVar> type_params;
   /*!
    * \brief potential constraint the type need to obey
-   * \note this field is reserved for futher purposes.
+   * \note this field is reserved for further purposes.
    */
   Array<TypeConstraint> type_constraints;
 

include/tvm/meta_schedule/cost_model.h

@@ -105,7 +105,7 @@ class PyCostModelNode : public CostModelNode {
    * \brief Predict the running results of given measure candidates.
    * \param context The tuning context.
    * \param candidates The measure candidates.
-   * \param p_addr The address to save the the estimated running results.
+   * \param p_addr The address to save the estimated running results.
    */
   using FPredict = runtime::TypedPackedFunc<void(const TuneContext&, const Array<MeasureCandidate>&,
                                                  void* p_addr)>;

include/tvm/relay/transform.h

@@ -125,7 +125,7 @@ TVM_DLL Pass FoldConstant(bool fold_qnn = false);
 TVM_DLL Pass SplitArgs(int max_function_args);
 
 /*!
- * \brief Fuse operations into expr into seperate functions.
+ * \brief Fuse operations into expr into separate functions.
  *
  * \param fuse_opt_level Optimization level. If it is -1 it will be inferred from pass context.
  *

include/tvm/runtime/c_backend_api.h

@@ -40,7 +40,7 @@ extern "C" {
  * \param args The arguments
  * \param type_codes The type codes of the arguments
  * \param num_args Number of arguments.
- * \param out_ret_value The output value of the the return value.
+ * \param out_ret_value The output value of the return value.
  * \param out_ret_tcode The output type code of the return value.
  * \param resource_handle Pointer to associated resource.
  *

include/tvm/runtime/crt/module.h

@@ -42,7 +42,7 @@ typedef struct TVMModule {
 /*!
  * \brief Create a new module handle from the given TVMModule instance.
  * \param mod The module instance to register.
- * \param out_handle Pointer to recieve the newly-minted handle for this module.
+ * \param out_handle Pointer to receive the newly-minted handle for this module.
  * \return 0 on success, non-zero on error.
  */
 int TVMModCreateFromCModule(const TVMModule* mod, TVMModuleHandle* out_handle);

include/tvm/runtime/vm/vm.h

@@ -334,7 +334,7 @@ class TVM_DLL VirtualMachine : public runtime::ModuleNode {
   /*!
    * \brief Set one input tensor with given index to set of input tensors if need copy to given
    * device. \param tensors the input tensors set (destination) \param tensor some tensor (not
-   * neccessary DLTensor). \param index The input tensor index. \param dev device to copy if need.
+   * necessary DLTensor). \param index The input tensor index. \param dev device to copy if need.
    */
   void SetInputTensorWithIndex(std::vector<ObjectRef>& tensors,  // NOLINT(*)
                                const TVMArgValue& tensor, int index, Device dev);

include/tvm/target/target.h

@@ -49,7 +49,7 @@ class TargetNode : public Object {
   TargetKind kind;
   /*! \brief Target host information, must be Target type */
   Optional<ObjectRef> host;
-  /*! \brief Tag of the the target, can be empty */
+  /*! \brief Tag of the target, can be empty */
   String tag;
   /*! \brief Keys for this target */
   Array<String> keys;

include/tvm/te/schedule.h

@@ -364,7 +364,7 @@ class Schedule : public ObjectRef {
                             const Array<Operation>& readers);
   /*!
    * \brief Create a cache write tensor for producing tensor.
-   *  The the tensor will take over body of original tensor op.
+   *  The tensor will take over body of original tensor op.
    *
    *  This function can be used to do data layout transformation.
    *  If there is a split/fuse/reorder on the data parallel axis of tensor
@@ -381,7 +381,7 @@ class Schedule : public ObjectRef {
   TVM_DLL Array<Tensor> cache_write(const Array<Tensor>& tensor, const std::string& scope);
   /*!
    * \brief Create a cache write tensor for producing tensor.
-   *  The the tensor will take over body of original tensor op.
+   *  The tensor will take over body of original tensor op.
    *
    *  This function can be used to do data layout transformation.
    *  If there is a split/fuse/reorder on the data parallel axis of tensor

include/tvm/te/schedule_pass.h

@@ -95,7 +95,7 @@ Stmt ScheduleOps(Schedule s, Map<IterVar, Range> dom_map, bool debug_keep_trivia
  *  Perform this translation before running any TIR optimizations.
  *
  *  List of actions taken by the function:
- *  - Remove occurences of te::Tensor, te::Operation in the IR
+ *  - Remove occurrences of te::Tensor, te::Operation in the IR
  *    and replace them by corresponding IR nodes via tir::Buffer.
  *  - Add annotation of extern buffers using the buffer_map field
  *    in the PrimFunc type.

include/tvm/tir/op.h

@@ -537,7 +537,7 @@ TVM_DLL PrimExpr isfinite(PrimExpr x, Span span = Span());
 TVM_DLL PrimExpr isinf(PrimExpr x, Span span = Span());
 
 /*!
- * \brief sum of of source expression over axis
+ * \brief sum of source expression over axis
  * \param source The source expression.
  * \param axis List of iteration variables that will be used for reduction.
  * \param init The value with which to initialize the output.
@@ -548,7 +548,7 @@ TVM_DLL PrimExpr sum(PrimExpr source, Array<tir::IterVar> axis, Array<PrimExpr>
                      Span span = Span());
 
 /*!
- * \brief logical And of of source expression over axis
+ * \brief logical And of source expression over axis
  * \param source The source expression.
  * \param axis List of iteration variables that will be used for reduction.
  * \param init The value with which to initialize the output.
@@ -558,7 +558,7 @@ TVM_DLL PrimExpr all(PrimExpr source, Array<tir::IterVar> axis, Array<PrimExpr>
                      Span span = Span());
 
 /*!
- * \brief logical Or of of source expression over axis
+ * \brief logical Or of source expression over axis
  * \param source The source expression.
  * \param axis List of iteration variables that will be used for reduction.
  * \param init The value with which to initialize the output.
@@ -569,7 +569,7 @@ TVM_DLL PrimExpr any(PrimExpr source, Array<tir::IterVar> axis, Array<PrimExpr>
                      Span span = Span());
 
 /*!
- * \brief max of of source expression over axis
+ * \brief max of source expression over axis
  * \param source The source expression.
  * \param axis List of iteration variables that will be used for reduction.
  * \param init The value with which to initialize the output.
@@ -580,7 +580,7 @@ TVM_DLL PrimExpr max(PrimExpr source, Array<tir::IterVar> axis, Array<PrimExpr>
                      Span span = Span());
 
 /*!
- * \brief max of of source expression over axis
+ * \brief max of source expression over axis
  * \param source The source expression.
  * \param axis List of iteration variables that will be used for reduction.
  * \param init The value with which to initialize the output.
@@ -591,7 +591,7 @@ TVM_DLL PrimExpr min(PrimExpr source, Array<tir::IterVar> axis, Array<PrimExpr>
                      Span span = Span());
 
 /*!
- * \brief product of of source expression over axis
+ * \brief product of source expression over axis
  * \param source The source expression.
  * \param axis List of iteration variables that will be used for reduction.
  * \param init The value with which to initialize the output.

include/tvm/topi/transform.h

@@ -641,7 +641,7 @@ inline Array<Tensor> split(const Tensor& x, Array<PrimExpr> split_indices, int a
  *
  * \param x The input tensor
  * \param begin The indices to begin with in the slicing
- * \param end Indicies indicating end of the slice
+ * \param end Indices indicating end of the slice
  * \param strides Specifies the stride values, it can be negative
  * in that case, the input tensor will be reversed in that particular axis
  * \param name The name of the operation
@@ -698,7 +698,7 @@ inline Tensor dynamic_strided_slice(const Tensor& x, const Array<PrimExpr>& begi
  *
  * \param x The input tensor
  * \param begin The indices to begin with in the slicing
- * \param end Indicies indicating end of the slice
+ * \param end Indices indicating end of the slice
  * \param strides Specifies the stride values, it can be negative
  * in that case, the input tensor will be reversed in that particular axis
  * \param name The name of the operation
@@ -729,7 +729,7 @@ inline te::Tensor dynamic_strided_slice(const te::Tensor& x, const te::Tensor& b
  *
  * \param ishape The input tensor shape
  * \param begin The indices to begin with in the slicing
- * \param end Indicies indicating end of the slice
+ * \param end Indices indicating end of the slice
  * \param strides Specifies the stride values, it can be negative
  * in that case, the input tensor will be reversed in that particular axis
  * \param axes Axes along which slicing is applied. When it is specified, the length of begin, end,
@@ -755,7 +755,7 @@ inline Array<PrimExpr> StridedSliceOutputShape(
  *
  * \param x The input tensor
  * \param begin The indices to begin with in the slicing
- * \param end Indicies indicating end of the slice
+ * \param end Indices indicating end of the slice
  * \param strides Specifies the stride values, it can be negative
  * in that case, the input tensor will be reversed in that particular axis
  * \param axes Axes along which slicing is applied. When it is specified, the length of begin, end,
@@ -803,7 +803,7 @@ inline Tensor strided_slice_with_axes(const Tensor& x, const Array<Integer>& beg
  *
  * \param x The input tensor
  * \param begin The indices to begin with in the slicing
- * \param end Indicies indicating end of the slice
+ * \param end Indices indicating end of the slice
  * \param strides Specifies the stride values, it can be negative
  * in that case, the input tensor will be reversed in that particular axis
  * \param slice_mode Specifies the slice mode

python/tvm/_ffi/_ctypes/packed_func.py

@@ -39,7 +39,7 @@
 
 
 def _ctypes_free_resource(rhandle):
-    """callback to free resources when it it not needed."""
+    """callback to free resources when it is not needed."""
     pyobj = ctypes.cast(rhandle, ctypes.py_object)
     ctypes.pythonapi.Py_DecRef(pyobj)
 

python/tvm/_ffi/runtime_ctypes.py

@@ -283,7 +283,7 @@ def max_threads_per_block(self):
     def warp_size(self):
         """Number of threads that execute concurrently.
 
-        Returns device value for for cuda, rocm, and vulkan.  Returns
+        Returns device value for cuda, rocm, and vulkan.  Returns
         1 for metal and opencl devices, regardless of the physical
         device.  Returns remote device value for RPC devices.  Returns
         None for all other devices.

python/tvm/contrib/cutlass/build.py

@@ -324,7 +324,7 @@ def tune_cutlass_kernels(
 
     split_k_slices : list of int
         Split factor candidates for split-K GEMM. If split-K > 1, the GEMM K-loop is computed in
-        parallel accross split-K blocks, and a seperate global reduction kernel is launched to
+        parallel across split-K blocks, and a separate global reduction kernel is launched to
         accumulate partial reductions. The profiler will pick the best split-k factor from the
         given candidate list. Note that the larger split-K factor requires a larger workspace.
         Currently, parallel split-k has been tested only for wgrad. For GEMM and other conv2d

python/tvm/contrib/ethosu/cascader/device_config.py

@@ -155,7 +155,7 @@ def _get_output_cycles(
         ifm_dtype: str
             Datatype of the Input Feature Map tensor (IFM)
         ofm_dtype: str
-            Datatype of the Ouput Feature Map tensor (OFM)
+            Datatype of the Output Feature Map tensor (OFM)
         activation : str
             The activation function to use.
                 "NONE" - no activation function.

python/tvm/contrib/pipeline_executor_build.py

@@ -586,7 +586,7 @@ def get_config(self):
                             dep_item["input_name"] = dname
                         dep_conf.append(dep_item)
 
-                # The value of ouput_idx start from 0.
+                # The value of output_idx start from 0.
                 output["output_idx"] = int(binding.name)
                 output["dependencies"] = dep_conf
                 output_conf.append(output)

python/tvm/driver/tvmc/micro.py

@@ -266,7 +266,7 @@ def create_project_handler(args):
     try:
         project.generate_project_from_mlf(template_dir, project_dir, mlf_path, options)
     except ServerError as error:
-        print("The following error occured on the Project API server side: \n", error)
+        print("The following error occurred on the Project API server side: \n", error)
         sys.exit(1)
 
 
@@ -292,7 +292,7 @@ def build_handler(args):
         prj = project.GeneratedProject.from_directory(project_dir, options=options)
         prj.build()
     except ServerError as error:
-        print("The following error occured on the Project API server side: ", error)
+        print("The following error occurred on the Project API server side: ", error)
         sys.exit(1)
 
 
@@ -310,5 +310,5 @@ def flash_handler(args):
         prj = project.GeneratedProject.from_directory(project_dir, options=options)
         prj.flash()
     except ServerError as error:
-        print("The following error occured on the Project API server side: ", error)
+        print("The following error occurred on the Project API server side: ", error)
         sys.exit(1)

python/tvm/error.py

@@ -17,7 +17,7 @@
 """Structured error classes in TVM.
 
 Each error class takes an error message as its input.
-See the example sections for for suggested message conventions.
+See the example sections for suggested message conventions.
 To make the code more readable, we recommended developers to
 copy the examples and raise errors with the same message convention.
 

python/tvm/ir/base.py

@@ -294,7 +294,7 @@ def structural_hash(node, map_free_vars=False):
 
     map_free_vars : bool
         If map_free_vars is set to true, we will hash free variables
-        by the order of their occurences. Otherwise, we will hash by
+        by the order of their occurrences. Otherwise, we will hash by
         their in-memory pointer address.
 
     Return

python/tvm/ir/diagnostics/__init__.py

@@ -29,7 +29,7 @@
 
 def get_renderer():
     """
-    Get the the diagnostic renderer.
+    Get the diagnostic renderer.
 
     Returns
     -------

python/tvm/relay/backend/contrib/ethosu/legalize.py

@@ -974,7 +974,7 @@ def __init__(self):
 
 
 class MeanRewriter(DFPatternCallback):
-    """Convert ethosu.mean composite functions to to an equivalent legalization:
+    """Convert ethosu.mean composite functions to an equivalent legalization:
     - Case 1 (axis == [1, 2] and keepsdims == True):
         ethosu_depthwise_conv2d + ethosu_binary_elementwise
     - Case 2 (ifm qparams == ofm qparams): ethosu_pooling