MCPcopy Create free account

hub / github.com/pytorch/executorch / functions

Functions28,493 in github.com/pytorch/executorch

↓ 81 callersMethodreset
(self)
backends/mlx/llm/cache.py:537
↓ 80 callersMethodexecute
( self, custom_runner_cmd=None, method_index=0, output_callback: Optional[Call
backends/qualcomm/export_utils.py:428
↓ 79 callersMethodchange_args
Updates the args to the given stage id.
backends/arm/test/tester/test_pipeline.py:350
↓ 79 callersFunctionrequire_kwargs
Validate that only allowed kwargs are present. Raises ValueError if unexpected kwargs are found. Args: kwargs: The handler kwar
backends/mlx/ops.py:246
↓ 78 callersMethodargs
(self, node: Node)
backends/mlx/builder/program_builder.py:194
↓ 78 callersMethodset_global
(self, quantization_config: QuantizationConfig)
backends/xnnpack/quantizer/xnnpack_quantizer.py:336
↓ 75 callersMethodfind_nodes
(self, model: GraphModule)
backends/cortex_m/quantizer/node_finders.py:43
↓ 73 callersMethodadd_stage_after
Adds a stage after the given stage id.
backends/arm/test/tester/test_pipeline.py:277
↓ 72 callersFunctionget_symmetric_a16w8_quantization_config
16A8W quantization config: 16-bit activations, 8-bit weights. This configuration provides better accuracy than 8A8W while maintaining reasona
backends/arm/quantizer/arm_quantizer.py:298
↓ 71 callersMethodrun_method
Run a specific method with the given inputs. Args: method_name: Name of the method to run, defaults to "forward"
export/export.py:616
↓ 70 callersMethodexport
(self)
examples/models/llava/export_llava.py:57
↓ 69 callersMethodquantize
(self, quantize_stage: Optional[Quantize] = None)
backends/test/harness/tester.py:209
↓ 68 callersMethodcheck
(self, input: List[str])
backends/test/harness/tester.py:271
↓ 67 callersFunctiongenerate_qnn_executorch_compiler_spec
Helper function generating compiler specs for Qualcomm AI Engine Direct Args: soc_model: The SoC you plan to run the compiled model.
backends/qualcomm/utils/utils.py:1134
↓ 66 callersFunctiongenerate_htp_compiler_spec
Helper function generating backend options for QNN HTP Args: use_fp16: If true, the model is compiled to QNN HTP fp16 runtime.
backends/qualcomm/utils/utils.py:1017
↓ 65 callersMethodcheck
Runs various checks on the given graph module to make sure it contains the needed data for passes. Some checks that need to
exir/pass_manager.py:57
↓ 65 callersMethodto_int_or_vid
(self, v: Union[int, Slot])
backends/mlx/builder/program_builder.py:301
↓ 65 callersMethodupdate
Update cache with new K/V states and return FULL cache. This method follows the same signature as examples/models/llama/attention.py
backends/mlx/llm/cache.py:181
↓ 64 callersFunction_is_annotated
Given a list of nodes (that represents an operator pattern), return True if any of the node is annotated, otherwise return False
backends/qualcomm/quantizer/rules.py:37
↓ 64 callersMethodpush
( # noqa: C901 self, inputs=None, files=None, backends: Optional[Set[QnnExecu
backends/qualcomm/export_utils.py:366
↓ 63 callersFunctionsingle_op_builder
Create a graph module with a single op. Args: placeholders: Placeholders to be used as inputs to the GraphModule. op: The op to b
backends/test/graph_builder.py:116
↓ 61 callersMethodpull
(self, host_output_path, device_output_path=None, callback=None)
backends/qualcomm/export_utils.py:478
↓ 60 callersMethod__init__
(self)
backends/nxp/tests/models.py:850
↓ 59 callersMethodto_edge
(self, to_edge_stage: Optional[ToEdge] = None)
backends/test/harness/tester.py:224
↓ 58 callersFunctioncheck_or_raise
Raises runtime error if condition is false, with the given error message Args: condition: boolean condition to check err: er
backends/xnnpack/utils/utils.py:31
↓ 58 callersMethodencode
(self, src_tokens)
exir/tests/transformer.py:97
↓ 58 callersMethodput
(self, key: str, value: List[NormalizedConstraints])
backends/qualcomm/quantizer/validators.py:77
↓ 58 callersMethodsupport_extension
Return True if an extension is supported and enabled. Args: extension (str): Extension name (for example ``int4``, ``bf16``).
backends/arm/tosa/specification.py:355
↓ 58 callersMethodwrite_to_file
Write the Cord to a file.
exir/_serialize/_cord.py:46
↓ 57 callersMethoddevice
(self)
examples/models/llama/evaluate/eager_eval.py:66
↓ 57 callersMethodexported_program
(self)
backends/arm/test/tester/arm_tester.py:262
↓ 57 callersMethodload_program
Loads an ExecuTorch program from a PTE binary. Args: data: The binary program data to load. Can be a file path (str or Path),
runtime/__init__.py:285
↓ 56 callersFunctiongen_samsung_backend_compile_spec
A function to generate an ExecuTorch binary for Samsung Backend. Attributes: chipset (str): chipset name in SamsungChipset. For exam
backends/samsung/serialization/compile_options.py:62
↓ 56 callersMethodrequired_envs
(self, conditions=None)
backends/qualcomm/tests/test_qnn_delegate.py:9453
↓ 55 callersMethodname
(self)
codegen/api/types/signatures.py:44
↓ 54 callersMethoddtype
(self)
exir/tensor.py:222
↓ 54 callersMethodexecute
Runs the specified method of this module with the specified arguments. @param methodName name of the ExecuTorch method to run. @param inputs argument
extension/android/executorch_android/src/main/java/org/pytorch/executorch/Module.java:132
↓ 54 callersFunctionget_backend_type
(backend: str)
backends/qualcomm/export_utils.py:691
↓ 54 callersMethodsave
Serialize and save the ETRecord to the specified path for use in Inspector. The ETRecord should contains at least edge dialect progra
devtools/etrecord/_etrecord.py:138
↓ 54 callersFunctionsetup_common_args_and_variables
()
backends/qualcomm/export_utils.py:708
↓ 53 callersMethodindex
(self, item)
backends/nxp/backend/ir/tflite_generator/meta/meta.py:76
↓ 53 callersMethodload_config
config (Union[argparse.Namespace, str]): Accepts either a parser generated from setup_common_args_and_variables() or a json file.
backends/qualcomm/export_utils.py:165
↓ 53 callersFunctionmake_output_dir
(path: str)
examples/qualcomm/utils.py:91
↓ 53 callersMethodpropose_partitions
Propose partitions for the graph module based on node groups and operator support. Returns: A list of proposed partition
exir/backend/canonical_partitioners/group_partitioner.py:337
↓ 53 callersMethodtransform
Transforms the program according to the provided passes. Args: passes: This param can be one of: 1) a li
exir/program/_program.py:1605
↓ 52 callersMethodget
(self, index: int)
backends/nxp/backend/ir/tflite_generator/meta/meta.py:82
↓ 52 callersMethodmax
(self)
devtools/inspector/_inspector.py:307
↓ 51 callersFunctionget_symmetric_quantization_config
Create symmetric quantization config for activations and weights. Activations use an affine qscheme; "symmetric" refers to the weight quantiz
backends/arm/quantizer/arm_quantizer.py:111
↓ 51 callersMethodsave
Saves the serialized ExecuTorch binary to the file at `path`.
exir/program/_program.py:2030
↓ 51 callersFunctionto_backend
A generic function the dispatch happens on the type of the first argument. There are currently to overloaded to_backend function: Note: Pyth
exir/backend/backend_api.py:44
↓ 50 callersMethodadd_default_cmds
(self, cmds)
backends/qualcomm/tests/utils.py:311
↓ 50 callersMethodcall
(self, graph_module: torch.fx.GraphModule)
backends/cadence/aot/tests/test_memory_passes.py:1154
↓ 49 callersFunctiondead_code_elimination_pass
(graph_module: torch.fx.GraphModule)
exir/passes/__init__.py:482
↓ 49 callersMethodpow
(lhs: _ExactValues, rhs: _ExactValues)
backends/arm/_passes/symbolic_value_range.py:84
↓ 48 callersMethodget_eager_model
(self)
examples/models/edsr/model.py:20
↓ 48 callersMethodis_available
Checks if a specific backend is available in the runtime. Args: backend_name: The name of the backend to check (e.g., "XnnpackBac
runtime/__init__.py:234
↓ 48 callersMethodrun
( self, graph_module: torch.fx.GraphModule, graph_signature: Optional[ExportGraphSigna
exir/tests/test_memory_planning.py:247
↓ 48 callersFunctionvalidate
Validate that two graph modules produce numerically equivalent outputs. Args: original: The original graph module before the pass
backends/cadence/aot/tests/test_replace_ops_passes.py:62
↓ 47 callersMethodcheckArgument
(boolean expression, String errorMessage, Object... args)
extension/android/executorch_android/src/main/java/org/pytorch/executorch/Tensor.java:1039
↓ 47 callersMethodis_channels_first
(self)
backends/nxp/backend/data_format.py:35
↓ 47 callersMethodmake_tmp_slot
(self)
backends/mlx/builder/slot_manager.py:131
↓ 46 callersMethod__init__
(self)
exir/emit/test/test_emit.py:662
↓ 46 callersMethod_check
(self, model, example_inputs, key, before_count, after_count)
extension/llm/export/test_export_passes.py:19
↓ 46 callersFunction_trace_and_lower_to_edge_ops
(f: Callable)
exir/passes/_quant_patterns_and_replacements.py:597
↓ 45 callersMethodsupport_integer
Return True if the ``INT`` profile is present.
backends/arm/tosa/specification.py:347
↓ 45 callersMethodupdate
Update KV cache with the new ``k_val``, ``v_val`` and return the updated cache. Note: When updating the KV cache, it is assumed t
extension/llm/modules/kv_cache.py:66
↓ 44 callersMethodSize
(self)
backends/nxp/backend/ir/lib/tflite/Buffer.py:76
↓ 44 callersMethodget_artifact
(self, stage: Optional[StageType] = None)
backends/test/harness/tester.py:267
↓ 44 callersMethodto_edge_transform_and_lower
( self, edge_compile_config: Optional[EdgeCompileConfig] = None, )
backends/samsung/test/tester/samsung_tester.py:117
↓ 43 callersFunctionbuild_executorch_binary
A function to generate an ExecuTorch binary for Qualcomm platforms. Attributes: model (torch.nn.Module): The model to be converted i
backends/qualcomm/export_utils.py:502
↓ 43 callersMethoddefine_op
( self, name, type, input_ids: List[int], output_ids: List[int],
backends/samsung/serialization/enn_graph_schema.py:34
↓ 42 callersFunctionmake_quantizer
( quant_dtype: Optional[QuantDtype] = QuantDtype.use_8a8w, custom_annotations=(), per_channel_conv
backends/qualcomm/export_utils.py:652
↓ 41 callersMethodcond
(i, h, h_accum)
exir/tests/test_passes.py:902
↓ 41 callersMethoddefine_nodes_tensor_inputs_outputs
( self, node: torch.fx.Node, xnn_graph: XNNGraph, vals_to_ids: Dict[torch.fx.N
backends/xnnpack/operators/node_visitor.py:652
↓ 40 callersMethodadd_named_data
Adds a named blob to the NamedDataStore. Args: key (str): key associated with the data. data (Union[bytes, to
exir/_serialize/_named_data_store.py:162
↓ 40 callersFunctioncopy_meta
(meta: Dict, callback=None)
backends/qualcomm/_passes/utils.py:16
↓ 40 callersMethodinsert
(self, index: int, item)
backends/nxp/backend/ir/tflite_generator/meta/meta.py:73
↓ 40 callersMethodpack
Stack per-expert weights into 3D buffers and delete the ModuleList. Must be called after quantization (if any) and before forward/export.
backends/mlx/llm/switch.py:88
↓ 39 callersMethodSize
(self)
exir/_serialize/generated/executorch_flatbuffer/DataSegment.py:40
↓ 39 callersFunction_is_float_tensor
Check if the node's tensor is a float tensor, so that we can skip quantization for the node since observers only works with float Tensors
backends/qualcomm/quantizer/rules.py:51
↓ 39 callersFunctioncreate_program
Returns an executorch program based on ModuleAdd, along with inputs.
extension/pybindings/test/make_test.py:151
↓ 39 callersMethoddefine_tensor
Defines a tensor value into the MPSGraph serialization schema Args: node (torch.fx.Node): EdgeIR tensor to define into mps_graph
backends/apple/mps/operators/node_visitor.py:71
↓ 39 callersFunctionget_shape
Return the shape of the tensor corresponding to node.
backends/transforms/permute_pass_utils.py:41
↓ 39 callersMethodsupport_float
Return True if the ``FP`` profile is present.
backends/arm/tosa/specification.py:351
↓ 39 callersFunctionvalidate_valid_dtype
Validate that one or more tensors have allowed dtypes. This function checks whether the ``dtype`` attribute of the provided tensor(s) is one
backends/arm/operators/operator_validation_utils.py:97
↓ 38 callersMethoddefine_tensor
Defines an tensor value into the XNNGraph Args: tensor: EdgeIR Tensor that is being defined into xnn_graph x
backends/xnnpack/operators/node_visitor.py:376
↓ 38 callersFunctionquantize_weight
Quantize ``weight`` to a torchao tensor subclass. Returns ``Int4Tensor`` for 4-bit or ``IntxUnpackedToInt8Tensor`` for 8-bit.
examples/models/gemma4_31b/quant/quantize.py:222
↓ 37 callersMethodremove
(self, item)
backends/nxp/backend/ir/tflite_generator/meta/meta.py:79
↓ 36 callersMethodget_inputs
(self)
backends/arm/test/ops/test_eq.py:42
↓ 36 callersMethodrun_passes
(self, run_passes_stage: Optional[RunPasses] = None)
backends/test/harness/tester.py:241
↓ 35 callersMethodallowed_to_transform
(self, meta: NodeMetadata | dict[str, Any])
backends/arm/_passes/arm_pass.py:30
↓ 35 callersMethodcopy
(self)
exir/pass_base.py:115
↓ 35 callersFunctionget_input_node
(node: torch.fx.Node, input_index: int)
backends/apple/mps/utils/mps_utils.py:14
↓ 34 callersMethodappend
(self, op_yaml_info: EdgeOpYamlInfo)
exir/dialects/edge/spec/gen.py:337
↓ 34 callersMethodcheck_graph_closeness
(self, graph_a, graph_b)
devtools/etrecord/tests/etrecord_test.py:169
↓ 34 callersFunctionget_param_tensor
( exp_prog: ExportedProgram, node: torch.fx.Node )
backends/transforms/utils.py:78
↓ 34 callersMethodmatch
Try to match the `Op` with a real TFLite Operator. If the match is successful, add new mappings for matched tensors into the tensor_map`.
backends/nxp/backend/ir/neutron_ir_post_processing/pattern_matcher.py:83
↓ 34 callersFunctionsave_tensors
(tensors, prefix, artifact_dir)
examples/samsung/utils.py:8
↓ 33 callersFunctioncount_node
Count the number of nodes with target `target` in the graph.
backends/transforms/test/test_permute_optimization_passes.py:33
↓ 33 callersMethodis_supported_fused_activation__aten
Node operator is supported fused activation on Neutron for Linear and Conv2D.
backends/nxp/backend/neutron_target_spec.py:27
← previousnext →101–200 of 28,493, ranked by callers