MCPcopy Create free account

hub / github.com/pytorch/executorch / functions

Functions28,493 in github.com/pytorch/executorch

↓ 13 callersFunctionget_node_arg
Help-function for getting a value from node.args/ kwargs, three cases: 1. By position in node.args - Returns arg at given position or default_val
backends/arm/_passes/arm_pass_utils.py:306
↓ 13 callersFunctionget_test_program
()
exir/tests/common.py:31
↓ 13 callersMethodis_nchw_node
(node: torch.fx.Node)
backends/xnnpack/_passes/channels_last_tagged_reshape_pass.py:126
↓ 13 callersFunctionis_param_node
Check if the given node is a parameter within the exported program
backends/vulkan/utils.py:185
↓ 13 callersMethodlower_unary_module_and_test_output
(self, module)
backends/vulkan/test/test_vulkan_delegate.py:466
↓ 13 callersMethodmake_or_get_constant
Creates an extra constant outside of the ExportedProgram state_dict. Ops can use this to add constants during build that do not exist
backends/mlx/builder/program_builder.py:245
↓ 13 callersFunctionmaketest
r"""Returns a TestCase method to test the provided module class and method. Args: module_cls: The subclass of nn.Module to export.
test/end2end/test_end2end.py:457
↓ 13 callersMethodop_code_index_for_op_type
Return the index to the 'operator_codes' vector in the TFLite model for the operator with given 'op_type' and 'version'. If correspon
backends/nxp/backend/ir/converter/builder/model_builder.py:744
↓ 13 callersMethodregister
Decorator to register a handler for one or more op targets.
backends/mlx/builder/op_registry.py:77
↓ 13 callersMethodrun
(self)
examples/qualcomm/oss_scripts/llama/decoder_runtime_evaluator.py:233
↓ 13 callersFunctionrun_model
( model: str, model_type: MODEL_TYPE = MODEL_TYPE.EXIR_DEFAULT_MODEL, use_fp16: bool = False,
backends/apple/mps/test/test_mps.py:61
↓ 13 callersFunctionto_model_input_spec
( input_spec: Iterable[ModelInputSpec] | tuple[int, ...] | list[tuple[int, ...]] )
backends/nxp/tests/executorch_pipeline.py:100
↓ 13 callersMethodupdate
Updates the Trie with new tokens provided as arguments. Args: *args: Variable number of words to be added to the Trie.
examples/mediatek/aot_utils/llm_utils/tokenizers_/tokenization_utils.py:60
↓ 12 callersMethod__init__
(self)
backends/xnnpack/test/passes/test_channels_last_tagged_reshape.py:490
↓ 12 callersMethod__init__
(self, quantizers: List[Quantizer])
backends/cadence/aot/quantizer/quantizer.py:297
↓ 12 callersMethod_coreml_partitioner
(self, *, minimum_deployment_target=ct.target.iOS18)
backends/apple/coreml/test/test_torch_ops.py:30
↓ 12 callersFunction_is_annotated
Given a list of nodes (that represents an operator pattern), check if any of the node is annotated, return True if any of the node is ann
backends/xnnpack/quantizer/xnnpack_quantizer_utils.py:104
↓ 12 callersFunction_make_codebook_and_rotation
Precompute TQ4 constants.
backends/cuda/tests/test_tq4_sdpa.py:48
↓ 12 callersFunction_make_tensor_arg_node
Create a mock arg node that looks like a single tensor node.
backends/vulkan/test/test_vulkan_tensor_repr.py:62
↓ 12 callersMethod_run_and_compare
Helper to export, run, and compare outputs.
kernels/test/test_bitwise_shift.py:95
↓ 12 callersMethod_safe_remove_node
(self, node, graph)
backends/xnnpack/_passes/remove_redundant_copy_pass.py:18
↓ 12 callersFunctionaligned_size
Returns input_size padded up to the next whole multiple of alignment.
exir/_serialize/padding.py:37
↓ 12 callersMethodcall_operator
(self, op, args, kwargs, meta, updated=False)
backends/arm/_passes/decompose_meandim_pass.py:103
↓ 12 callersMethodcheckNotDestroyed
()
extension/android/executorch_android/src/main/java/org/pytorch/executorch/extension/llm/LlmModule.java:193
↓ 12 callersMethodcheckNotReentrant
()
extension/android/executorch_android/src/main/java/org/pytorch/executorch/extension/llm/LlmModule.java:197
↓ 12 callersFunctioncheck_annotation
(model: SingleOpModel)
backends/arm/test/quantizer/test_generic_annotater.py:39
↓ 12 callersFunctioncombo_pattern
Returns the cartesian product of the given pattern lists.
backends/arm/quantizer/quantizer_support.py:18
↓ 12 callersFunctioncompute_sqnr
(x: torch.Tensor, y: torch.Tensor)
examples/models/moshi/mimi/test_mimi.py:32
↓ 12 callersMethodconstrain_to_compatible_packed_dim
Return a new TensorRepSet containing only layouts from self whose PackedDimInfo is present in other's PackedDimInfo sets. If other is
backends/vulkan/utils.py:1062
↓ 12 callersMethodcreate_transpose_operator_before
Create a TFLite 'Transpose' operator before the 'before_operator'. The input of 'before_operator' at index 'on_input_index',
backends/nxp/backend/ir/converter/builder/model_builder.py:951
↓ 12 callersFunctiondeserialize
( artifact: export_serialize.SerializedArtifact, )
exir/serde/serialize.py:723
↓ 12 callersMethoddump_intermediate_artifacts_to
Sets a path for dumping intermediate results during such as tosa and pte. Args: output_path: Path to dump intermediate re
backends/arm/common/arm_compile_spec.py:308
↓ 12 callersMethodforward
(self, x, y)
extension/pybindings/test/make_test.py:22
↓ 12 callersMethodfrom_attention_mha
( cls, other: AttentionMHA, split_mha: bool = True, rms_norm_class=torch.nn.RM
examples/models/llama/static_attention.py:918
↓ 12 callersMethodfrom_inputs
( cls, tensor_node: torch.fx.Node, ep: ExportedProgram )
backends/xnnpack/operators/quant_params.py:349
↓ 12 callersFunctiongenerate_mask
( cache_size, valid_cache, input_length, valid_input, batch_size=1, mask_value=-100.0,
examples/mediatek/aot_utils/llm_utils/utils.py:294
↓ 12 callersMethodget_artifacts
(self)
export/stages.py:99
↓ 12 callersMethodget_context
(self, key: str, default: Any = None)
export/stages.py:45
↓ 12 callersMethodget_delegate_mapping
Returns: Union[Dict[int, Tuple[int]], Dict[str, Tuple[int]]]: A map of delegate debug identifier to a list of debu
exir/backend/utils.py:467
↓ 12 callersFunctionget_delegation_info
Util function to get the delegation information of the given graph module. Args: graph_module: The lowered graph module to get the d
devtools/backend_debug/delegation_info.py:111
↓ 12 callersFunctionget_embedding_replacement_filter
(has_nonzero_zero_point)
exir/passes/_quant_patterns_and_replacements.py:993
↓ 12 callersMethodget_inputs
(self)
backends/arm/test/ops/test_ne.py:45
↓ 12 callersFunctionis_tensor_node
Returns true if the given node produces a tensor value, or a collection of tensor values
backends/vulkan/utils.py:248
↓ 12 callersMethodmakeExecutorchException
(int errorCode, String details)
extension/android/executorch_android/src/main/java/org/pytorch/executorch/ExecutorchRuntimeException.java:189
↓ 12 callersFunctiontag_as_implicit_q_dq
(node: torch.fx.Node)
backends/xnnpack/utils/quant_utils.py:52
↓ 12 callersMethodtensor_for_name
Get an existing TFLite tensor with given 'name'. If such tensor does NOT exist, function will create and register a new tensor with s
backends/nxp/backend/ir/converter/builder/model_builder.py:805
↓ 12 callersMethodto_edge
( self, to_edge_stage: Optional[BaseStages.ToEdge] = None, # Keep config keyword-only
backends/arm/test/tester/arm_tester.py:419
↓ 12 callersFunctionto_edge_program
( model: nn.Module, input_spec: Iterable[ModelInputSpec] | tuple[int, ...] | list[tuple[int, ...]], )
backends/nxp/tests/executorch_pipeline.py:306
↓ 12 callersFunctionto_edge_transform_and_lower_to_enn
( module: torch.nn.Module, inputs: Tuple[torch.Tensor], custom_pass_config: List[PassType] = None,
backends/samsung/utils/export_utils.py:77
↓ 11 callersMethodGetRootAs
(cls, buf, offset=0)
backends/nxp/backend/ir/lib/tflite/Model.py:15
↓ 11 callersMethod__init__
(self)
exir/backend/test/test_to_backend_multi_method.py:136
↓ 11 callersMethod__init__
(self, config)
backends/arm/test/models/Qwen3_VL/test_qwen3_vl_layers.py:346
↓ 11 callersMethod__init__
(self, config, layer_idx)
examples/models/qwen3_5_moe/model.py:627
↓ 11 callersFunction_as_concrete_ints
(values: Sequence[int | torch.SymInt])
backends/arm/tosa/resize_utils.py:20
↓ 11 callersMethod_compare_outputs
(self, executorch_program, eager_program, example_inputs)
backends/apple/coreml/test/test_torch_ops.py:43
↓ 11 callersMethod_export_to_cuda_with_lower
Helper method to export a module to CUDA backend using to_edge_transform_and_lower. Args: module: The torch.nn.Module to export
backends/cuda/tests/test_cuda_export.py:28
↓ 11 callersMethod_fx
( self, kind: str, target: torch.fx.node.Target, args: Tuple[Argument, ...],
exir/pass_base.py:431
↓ 11 callersMethod_internal_assert_emitter
If pred is False, construct and raise a node specific error message.
exir/emit/_emitter.py:266
↓ 11 callersFunction_make_rescale
( graph: torch.fx.Graph, input_node: torch.fx.Node, output_dtype: torch.dtype, scale: float,
backends/arm/test/passes/test_rescale_optimization.py:104
↓ 11 callersMethod_rand_kv
(self, seq_len)
examples/models/llama/source_transformation/test_attention_sink.py:147
↓ 11 callersMethod_run_model_with_all_recipes
(self, model_name: str)
export/tests/test_target_recipes.py:456
↓ 11 callersFunction_sdpa_with_kv_cache_ref
(q, k, v, k_cache, v_cache, attn_mask, start_pos, seq_len)
extension/llm/custom_ops/test_sdpa_with_kv_cache.py:22
↓ 11 callersMethod_test_sdpa_common
( self, n_heads_kv, n_heads_q, head_dim, max_seq_len, start_po
extension/llm/custom_ops/test_quantized_sdpa.py:222
↓ 11 callersMethodadd_passes
(self, passes: Sequence[ExportPass | None])
backends/arm/_passes/arm_pass_manager.py:371
↓ 11 callersFunctionapply_algo
Recursively apply algo to graph_module and its submodules for control flow. Partitions specs by device type and device idx, and runs the mem
exir/memory_planning.py:1411
↓ 11 callersFunctionbuild_result
( result: TestResult, error: Exception | None = None )
backends/test/suite/runner.py:110
↓ 11 callersMethodcalculate_positions_and_update_indices
Calculate indices into k_cache, v_cache for placing k_val, v_val. Sink tokens (positions < sink_size) map to cache slots [0, sink_si
examples/models/llama/source_transformation/attention_sink.py:157
↓ 11 callersMethodcalculate_qparams
Calculates quantization parameters: quantized weight, quantization scale and quantization zero point. :param weight: FP weight to be
backends/openvino/quantizer/observers.py:64
↓ 11 callersMethodcall_operator
(self, op, args, kwargs, meta)
backends/arm/_passes/decompose_atan_pass.py:79
↓ 11 callersMethodcheck_fully_delegated
Helper to verify a program is fully delegated to CoreML
backends/apple/coreml/test/test_coreml_recipes.py:41
↓ 11 callersMethoddefine_custom_tensor_wrapper
( self, node_name: str, tensor_type: PyQnnManager.Qnn_TensorType_t, dtype: PyQ
backends/qualcomm/builders/node_visitor.py:429
↓ 11 callersMethodembed_tokens
(self, tokens: torch.Tensor)
examples/models/llava/model.py:159
↓ 11 callersMethodencode
Converts a string to a sequence of ids (integer), using the tokenizer and vocabulary. Same as doing `self.convert_tokens_to_ids(self.tokenize
examples/mediatek/aot_utils/llm_utils/tokenizers_/tokenization_utils_base.py:2783
↓ 11 callersFunctioneval_expr
Evaluate a symint to int. Returns None if symint's symoblic expr can not be evaluated to valid integer according to the hints.
exir/sym_util.py:17
↓ 11 callersMethodexecute_layer_test
( self, module: torch.nn.Module, sample_inputs: tuple[torch.Tensor], expected_
backends/openvino/tests/ops/base_openvino_op_test.py:20
↓ 11 callersFunctionflatbuffer_to_option
(flatbuffers: bytes)
backends/qualcomm/serialization/qc_schema_serialize.py:58
↓ 11 callersMethodforward
( self, tokens: torch.Tensor, input_pos: Optional[torch.Tensor] = None, )
examples/models/llama/runner/generation.py:84
↓ 11 callersMethodget
Returns the current contents of the named file.
exir/_serialize/_flatbuffer.py:141
↓ 11 callersMethodget_example_inputs
(self)
backends/samsung/test/ops/test_bmm.py:23
↓ 11 callersMethodget_example_inputs
(self)
backends/arm/test/ops/test_elu.py:81
↓ 11 callersMethodget_example_inputs
(self)
examples/samsung/scripts/mobilebert_finetune.py:35
↓ 11 callersFunctionget_exp_name
(config_path)
examples/mediatek/aot_utils/llm_utils/utils.py:424
↓ 11 callersMethodget_out_repset
(self, i: int)
backends/vulkan/utils.py:1492
↓ 11 callersFunctionget_platform
()
util/collect_env.py:378
↓ 11 callersMethodget_program
(self)
backends/test/program_builder.py:109
↓ 11 callersMethodget_relative_placement_source
( self, node: torch.fx.Node )
backends/cadence/aot/memory_constraints.py:94
↓ 11 callersMethodget_slot
( self, node_or_name: Union[Node, str] )
backends/mlx/builder/slot_manager.py:91
↓ 11 callersFunctionget_xnnpack_edge_compile_config
( skip_dim_order: bool = False, )
backends/xnnpack/utils/configs.py:14
↓ 11 callersMethodindex
(t, input_pos)
examples/models/llama/source_transformation/test_quantized_kv_cache.py:64
↓ 11 callersMethodinsert_passes_before
Register passes to be inserted before instances of target_pass_type. Insertions are deferred and applied via _apply_pass_insertions().
backends/arm/_passes/arm_pass_manager.py:275
↓ 11 callersFunctionis_depthwise_conv
Check whether a convolution is depthwise. Depthwise convolution has groups == in_channels with groups > 1. When groups == 1, it is always a r
backends/cadence/aot/utils.py:52
↓ 11 callersMethodis_enabled
(self, var: str, fallback: bool = False)
tools/cmake/cmake_cache.py:30
↓ 11 callersMethodis_quant_node
(node: torch.fx.Node)
backends/nxp/neutron_partitioner.py:95
↓ 11 callersFunctionmerge_decomposed_graph
( remap: Dict[str, torch.fx.Node], target_node: torch.fx.Node, target_graph: torch.fx.GraphModule,
backends/qualcomm/_passes/utils.py:140
↓ 11 callersFunctionnum_bytes_from_shape_and_dtype
Assume the tensor is a contiguous one.
exir/tensor.py:30
↓ 11 callersMethodoutput
Performs actions for the output node of a graph module. The output node of the top level entry point is handled by TopLevelEmitter. This func
exir/emit/_emitter.py:1733
↓ 11 callersFunctionpack_model
Pack a state dict into ``model`` using the given packers. Quantized weights (torchao tensor subclasses) are grouped by parent module and disp
examples/models/gemma4_31b/quant/pack.py:32
↓ 11 callersMethodplaceholder
Performs actions for the placeholder node of a graph module. The placeholder node of the top level entry point is handled by TopLevelEmitter.
exir/emit/_emitter.py:1711
↓ 11 callersFunctionread_file
Returns the contents of the given file.
exir/_serialize/test/test_flatbuffer.py:25
↓ 11 callersMethodrun_executorch_test
Export to ExecuTorch and compare with PyTorch reference.
kernels/portable/test/test_grid_sampler_2d_executorch.py:50
← previousnext →501–600 of 28,493, ranked by callers