MCPcopy Create free account

hub / github.com/pytorch/executorch / functions

Functions28,493 in github.com/pytorch/executorch

↓ 1,769 callersMethodrun
(self, artifact: ExecutorchProgramManager, inputs=None)
backends/arm/test/tester/serialize.py:52
↓ 1,382 callersMethodto
(self, *args, **kwargs)
examples/models/llama/experimental/subclass.py:343
↓ 1,183 callersMethodappend
Append a bytes or Cord to the current Cord.
exir/_serialize/_cord.py:35
↓ 1,144 callersMethodappend
(self, val: TensorRepr)
backends/vulkan/utils.py:855
↓ 1,021 callersMethod_test_op
(self, model, args, flow, generate_random_test_inputs=False)
backends/test/suite/operators/__init__.py:72
↓ 990 callersMethodones
Creates a new Tensor instance with given data-type and all elements initialized to one. @param shape Tensor shape @param dtype Tensor data-type
extension/android/executorch_android/src/main/java/org/pytorch/executorch/Tensor.java:401
↓ 790 callersMethodzeros
Creates a new Tensor instance with given data-type and all elements initialized to zero. @param shape Tensor shape @param dtype Tensor data-type
extension/android/executorch_android/src/main/java/org/pytorch/executorch/Tensor.java:442
↓ 752 callersMethodinfo
(node)
examples/qualcomm/oss_scripts/llama/wrappers/llm_wrappers.py:823
↓ 663 callersMethodget
Returns a list of values associated with the given TOSA specification. The specification is normalized to its canonical form, which m
backends/arm/tosa/specification.py:64
↓ 630 callersMethoditems
Get dict itmes.
examples/mediatek/aot_utils/llm_utils/tokenizers_/tokenization_utils_base.py:269
↓ 579 callersMethodcall_operator
( self, op, # pyre-ignore args: tuple[Argument, ...], kwargs: Optional[dict[s
backends/test/graph_builder.py:81
↓ 576 callersFunctionexport
Create and configure an ExportSession with the given parameters. This function provides a convenient way to create an ExportSession and
export/export.py:42
↓ 566 callersMethodkeys
Get dict keys.
examples/mediatek/aot_utils/llm_utils/tokenizers_/tokenization_utils_base.py:261
↓ 503 callersMethodexported_program
Returns the ExportedProgram specified by 'method_name'.
exir/program/_program.py:1597
↓ 502 callersMethodOffset
(self)
backends/nxp/backend/ir/lib/tflite/Buffer.py:67
↓ 450 callersMethodslot_to_tid
Convert a tensor Slot to a Tid, recording it for later remapping.
backends/mlx/builder/program_builder.py:286
↓ 443 callersMethodexport
(self, export_stage: Optional[Export] = None)
backends/test/harness/tester.py:215
↓ 443 callersFunctionto_edge
:func:`to_edge` constructs an EdgeProgramManager from a set of exported programs in ATen dialect. Upon construction those programs are transf
exir/program/_program.py:1451
↓ 398 callersMethodlower_module_and_test_output
( self, module: torch.nn.Module, sample_inputs: Tuple[torch.Tensor], expected_
backends/qualcomm/tests/utils.py:661
↓ 353 callersMethodmodule
(self)
devtools/inspector/tests/inspector_test.py:1309
↓ 338 callersMethodcheck_count
(self, input: Dict[Any, int])
backends/test/harness/tester.py:281
↓ 327 callersMethodget
()
exir/tracer.py:478
↓ 325 callersMethodregister_buffer
Add buffer to self.exported_program.state_dict.
backends/arm/_passes/insert_table_ops.py:140
↓ 317 callersFunctioncount_node
Count the number of nodes with target `target` in the graph.
backends/cadence/aot/pass_utils.py:136
↓ 317 callersMethodsplit
Will look for the words added to the trie within `text`. Output is the original string splitted along the boundaries of the words found.
examples/mediatek/aot_utils/llm_utils/tokenizers_/tokenization_utils.py:99
↓ 296 callersMethodto_executorch
(self, to_executorch_stage: Optional[ToExecutorch] = None)
backends/test/harness/tester.py:251
↓ 283 callersMethoddefault
(self, obj)
exir/serde/export_serialize.py:2312
↓ 276 callersMethodload
Loads a serialized ExecuTorch module from the specified path on the disk. @param modelPath path to file that contains the serialized ExecuTorch modul
extension/android/executorch_android/src/main/java/org/pytorch/executorch/Module.java:85
↓ 270 callersMethodexported_program
(self)
backends/xnnpack/_passes/xnnpack_pass.py:21
↓ 259 callersMethodplaceholder
( self, target: str, fake_tensor: Union[FakeTensor, torch.Tensor] )
backends/test/graph_builder.py:64
↓ 250 callersMethoddefine_tensor
Covert torch.Tensor to TensorWrapper Args: tensor_source_node: EdgeIR Node target_build_node: Current node t
backends/qualcomm/builders/node_visitor.py:464
↓ 244 callersMethodwrite
(self, text)
examples/mediatek/aot_utils/llm_utils/loggingprinter.py:19
↓ 243 callersFunctiongraph_contains_any_of_ops
(graph: Graph, ops: list)
backends/nxp/tests/executors.py:387
↓ 238 callersMethod__init__
(self, pos, neg)
backends/qualcomm/tests/models.py:26
↓ 237 callersMethodto
Send all values to device by calling `v.to(device)` (PyTorch only). Args: device (`str` or `torch.device`): The device to put the
examples/mediatek/aot_utils/llm_utils/tokenizers_/tokenization_utils_base.py:801
↓ 235 callersMethodget_tensor
Get tensor value/shape with axis_order
backends/qualcomm/builders/node_visitor.py:130
↓ 234 callersMethodinserting_before
Sets the point at which we will insert the graph.
exir/graph.py:44
↓ 230 callersFunctionramp_tensor
(start: float, end: float, shape: tuple[int, ...])
backends/cortex_m/test/tester.py:162
↓ 228 callersMethodrun_method_and_compare_outputs
( self, stage: Optional[StageType] = None, inputs: Optional[Tuple[torch.Tensor]] = Non
backends/test/harness/tester.py:316
↓ 224 callersMethodcall_function
Performs actions for the call_function node of a graph module. Dispatches based on 'target' and emits the corresponding function. 'call_funct
exir/emit/_emitter.py:1763
↓ 222 callersMethodclone
Create a clone of the KVCache.
extension/llm/modules/kv_cache.py:134
↓ 212 callersMethodget_qdq_module
( self, module: torch.nn.Module, inputs: Tuple[torch.Tensor], is_conv_per_chan
backends/qualcomm/tests/utils.py:752
↓ 210 callersMethodvalues
Get dict values.
examples/mediatek/aot_utils/llm_utils/tokenizers_/tokenization_utils_base.py:265
↓ 205 callersMethodpop_stage
Removes and returns the stage at postion pos.
backends/arm/test/tester/test_pipeline.py:249
↓ 195 callersMethodcheck_not
(self, input: List[str])
backends/test/harness/tester.py:276
↓ 195 callersMethodoutput
(self, results: list[ProxyValue])
backends/test/graph_builder.py:74
↓ 194 callersMethodemit
(self, op: OpNodeUnion)
backends/mlx/builder/program_builder.py:167
↓ 193 callersMethoddebug
(self)
exir/tensor.py:241
↓ 188 callersMethodpermute
(w, n_heads)
examples/models/llama/static_attention.py:1483
↓ 187 callersFunctioncreate_node
Adds a node to 'graph'. graph.inserting_before/after() should be used before the call to decide where to insert the node. If quantize is true
backends/arm/_passes/arm_pass_utils.py:141
↓ 186 callersFunctionto_quantized_edge_program
( model: torch.nn.Module, input_spec: Iterable[ModelInputSpec] | tuple[int, ...] | list[tuple[int, ...
backends/nxp/tests/executorch_pipeline.py:178
↓ 183 callersMethodget_graph_module
(self)
backends/test/graph_builder.py:78
↓ 174 callersMethodcreate_node
( self, graph: torch.fx.Graph, op_target: OpOverload, args: tuple = (),
backends/xnnpack/_passes/conv1d_unsqueeze_pass.py:32
↓ 173 callersMethodOffset
(self)
exir/_serialize/generated/executorch_flatbuffer/DataSegment.py:33
↓ 172 callersMethodto_edge_transform_and_lower
( self, to_edge_and_transform_stage: Optional[ToEdgeTransformAndLower] = None, generat
backends/test/harness/tester.py:230
↓ 170 callersMethoderase_node
Erases a ``Node`` from the ``Graph``. Throws an exception if there are still users of that node in the ``Graph``.
exir/graph.py:37
↓ 168 callersMethodlower_and_test_with_partitioner
( self, graph_module, example_inputs, func_name: str, use_fp16: bool =
backends/apple/mps/test/test_mps_utils.py:284
↓ 161 callersMethodto_executorch
Transforms the program to the ExecuTorch backend. Args: config: An optional argument used to provide greater control ove
exir/program/_program.py:1731
↓ 160 callersMethodadd
(self, node: torch.fx.Node, tosa_op: Any, tosa_op_id: ts.Op)
backends/arm/debug/schema.py:119
↓ 155 callersMethodcreate_from_string
Create a specification from a standard string format. Example: ``TOSA-1.00.0+INT+FP+int4+cf``. Args: repr (str): Standar
backends/arm/tosa/specification.py:187
↓ 151 callersMethodget
Gets the pass corresponding to the given name. If the pass is a function then it will directly return the callable function.
exir/passes/pass_registry.py:71
↓ 147 callersMethodserialize
(self, serialize_stage: Optional[Serialize] = None)
backends/test/harness/tester.py:256
↓ 133 callersMethodget
(self, key: str)
backends/qualcomm/quantizer/validators.py:74
↓ 132 callersMethodset_global
Set quantization_config for submodules not matched by other filters. Args: quantization_config (Optional[QuantizationConfig]): Co
backends/arm/quantizer/arm_quantizer.py:582
↓ 126 callersMethodcopy
(self)
backends/vulkan/utils.py:898
↓ 124 callersMethodget_node
Utility to skip dequantize node for frozen param
backends/qualcomm/builders/node_visitor.py:117
↓ 124 callersMethodmax
(self)
backends/arm/tosa/mapping.py:60
↓ 124 callersMethodreport_reject
Stores an RejectedPatternReport containing info about the rejected pattern.
backends/cortex_m/quantizer_reporter.py:157
↓ 123 callersMethodappend
(self, item)
backends/nxp/backend/ir/tflite_generator/meta/meta.py:70
↓ 119 callersMethodget
Gets the Runtime singleton.
runtime/__init__.py:272
↓ 118 callersFunctionto_edge_transform_and_lower
:func:`to_edge_transform_and_lower` constructs an EdgeProgramManager from a set of exported programs in ATen dialect. It differs fundamentall
exir/program/_program.py:1324
↓ 115 callersMethod_check
(self, out, ref, tol=0.15)
backends/cuda/tests/test_int4_dispatch.py:137
↓ 115 callersFunctionget_arg
Get the arg with kwarg_name of the node.
backends/transforms/permute_pass_utils.py:86
↓ 114 callersFunctionget_symmetric_quantization_config
( is_per_channel: bool = False, is_qat: bool = False, is_dynamic: bool = False, act_qmin: int
backends/xnnpack/quantizer/xnnpack_quantizer.py:106
↓ 112 callersMethodexport
Execute the full export process. This method orchestrates the export process with optional quantization: 1. (Optional) Apply
export/export.py:460
↓ 105 callersMethod__init__
(self)
backends/apple/mps/test/test_mps.py:244
↓ 104 callersMethodadd
(self, ngram: List[int])
examples/apple/coreml/llama/llama_transformer.py:460
↓ 104 callersMethodfrom_pretrained
Instantiate a [`~tokenization_utils_base.PreTrainedTokenizerBase`] from a predefined tokenizer. Args: pretrained_model_name_or_pa
examples/mediatek/aot_utils/llm_utils/tokenizers_/tokenization_utils_base.py:1899
↓ 104 callersMethodfrom_tensor
Creates a TensorStatistics object from a tensor.
backends/test/harness/error_statistics.py:30
↓ 101 callersMethodTensor
(long[] shape)
extension/android/executorch_android/src/main/java/org/pytorch/executorch/Tensor.java:473
↓ 101 callersMethoddefine_tensor
( self, node: torch.fx.Node, enn_graph: EnnGraph, vals_to_ids: Dict[torch.fx.N
backends/samsung/builders/node_visitor.py:37
↓ 101 callersMethodnumel
Returns the number of elements in this tensor.
extension/android/executorch_android/src/main/java/org/pytorch/executorch/Tensor.java:479
↓ 100 callersMethodcreate
(cls, **kwargs)
exir/serde/union.py:45
↓ 99 callersFunctionget_first_fake_tensor
Returns a FakeTensor from the meta field of 'node'. If the node contains many fake tensors, return the first one.
backends/arm/_passes/arm_pass_utils.py:283
↓ 98 callersMethod__init__
(self)
backends/vulkan/test/test_vulkan_delegate.py:269
↓ 98 callersFunctionconvert_run_compare
( edge_program: ExportedProgram, input_data, rtol=1.0e-5, atol=1.0e-8, save_models=False,
backends/nxp/tests/executors.py:310
↓ 97 callersMethoddecode
codes: (B, 1+acoustic_dim, T) -> (B, semantic_dim+acoustic_dim, T)
examples/models/voxtral_tts/model.py:1271
↓ 94 callersMethod__init__
( self, shape: Tuple[int, ...], dtype: torch.dtype, )
backends/mlx/test/test_ops.py:4185
↓ 93 callersMethodflatten
(self)
backends/nxp/backend/ir/converter/conversion/common.py:156
↓ 92 callersMethodlower_module_and_test_output
Helper testing function that takes a torch.nn.Module and lowers it to Vulkan with the given sample inputs. It then runs the lowered m
backends/vulkan/test/test_vulkan_delegate.py:224
↓ 90 callersFunctionget_input_node
(node: torch.fx.Node, input_index: int)
backends/xnnpack/utils/utils.py:57
↓ 88 callersFunction_build_module
( op_call: Callable[[torch.Tensor, torch.Tensor], torch.Tensor] )
backends/cortex_m/test/misc/test_portable_int8.py:36
↓ 88 callersMethodto_executorch
(self, artifact, pte_filename)
examples/qualcomm/oss_scripts/llm_utils/qnn_decoder_model_manager.py:325
↓ 86 callersMethodadd_pass
(self, pipeline_pass)
backends/arm/_passes/arm_pass_manager.py:383
↓ 86 callersMethodload_method
Loads a method from the program. Args: name: The name of the method to load. Returns: The loaded method.
runtime/__init__.py:183
↓ 85 callersMethodto_backend
Returns a semantically-equivalent program to the one given as input, but with portions of each program in the EdgeProgramManager targ
exir/program/_program.py:1680
↓ 84 callersMethodpad
Pad a single encoded input or a batch of encoded inputs up to predefined length. Padding side (left/right) padding token ids are defined at t
examples/mediatek/aot_utils/llm_utils/tokenizers_/tokenization_utils_base.py:3431
↓ 83 callersMethodkwargs
(self, node: Node)
backends/mlx/builder/program_builder.py:197
↓ 83 callersMethodmake_or_get_slot
Get or create a slot for a single-output node. Returns a single Slot. Use this for single-output ops (the common case). For
backends/mlx/builder/slot_manager.py:189
↓ 81 callersFunctionrequire_args
Validate that args count is within expected range. Raises ValueError if the count is outside the expected range. Args: args: Th
backends/mlx/ops.py:221
next →1–100 of 28,493, ranked by callers