Table of Contents
  • Docs >
  • Exporting to ExecuTorch Tutorial
Shortcuts
tutorials/export-to-executorch-tutorial

Note

Go to the end to download the full example code

Exporting to ExecuTorch Tutorial

Author: Angela Yi

ExecuTorch is a unified ML stack for lowering PyTorch models to edge devices. It introduces improved entry points to perform model, device, and/or use-case specific optimizations such as backend delegation, user-defined compiler transformations, default or user-defined memory planning, and more.

At a high level, the workflow looks as follows:

../_images/executorch_stack.png

In this tutorial, we will cover the APIs in the “Program preparation” steps to lower a PyTorch model to a format which can be loaded to device and run on the ExecuTorch runtime.

Prerequisites

To run this tutorial, you’ll first need to Set up your ExecuTorch environment.

Exporting a Model

Note: The Export APIs are still undergoing changes to align better with the longer term state of export. Please refer to this issue for more details.

The first step of lowering to ExecuTorch is to export the given model (any callable or torch.nn.Module) to a graph representation. This is done via torch.export, which takes in an torch.nn.Module, a tuple of positional arguments, optionally a dictionary of keyword arguments (not shown in the example), and a list of dynamic shapes (covered later).

import torch
from torch.export import export, ExportedProgram


class SimpleConv(torch.nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv = torch.nn.Conv2d(
            in_channels=3, out_channels=16, kernel_size=3, padding=1
        )
        self.relu = torch.nn.ReLU()

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        a = self.conv(x)
        return self.relu(a)


example_args = (torch.randn(1, 3, 256, 256),)
aten_dialect: ExportedProgram = export(SimpleConv(), example_args, strict=True)
print(aten_dialect)

ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, p_conv_weight: "f32[16, 3, 3, 3]", p_conv_bias: "f32[16]", x: "f32[1, 3, 256, 256]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:64 in forward, code: a = self.conv(x)
            conv2d: "f32[1, 16, 256, 256]" = torch.ops.aten.conv2d.default(x, p_conv_weight, p_conv_bias, [1, 1], [1, 1]);  x = p_conv_weight = p_conv_bias = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:65 in forward, code: return self.relu(a)
            relu: "f32[1, 16, 256, 256]" = torch.ops.aten.relu.default(conv2d);  conv2d = None
            return (relu,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.PARAMETER: 2>, arg=TensorArgument(name='p_conv_weight'), target='conv.weight', persistent=None), InputSpec(kind=<InputKind.PARAMETER: 2>, arg=TensorArgument(name='p_conv_bias'), target='conv.bias', persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='relu'), target=None)])
Range constraints: {}

The output of torch.export.export is a fully flattened graph (meaning the graph does not contain any module hierarchy, except in the case of control flow operators). Additionally, the graph is purely functional, meaning it does not contain operations with side effects such as mutations or aliasing.

More specifications about the result of torch.export can be found here .

The graph returned by torch.export only contains functional ATen operators (~2000 ops), which we will call the ATen Dialect.

Expressing Dynamism

By default, the exporting flow will trace the program assuming that all input shapes are static, so if we run the program with inputs shapes that are different than the ones we used while tracing, we will run into an error:

import traceback as tb


class Basic(torch.nn.Module):
    def __init__(self):
        super().__init__()

    def forward(self, x: torch.Tensor, y: torch.Tensor) -> torch.Tensor:
        return x + y


example_args = (torch.randn(3, 3), torch.randn(3, 3))
aten_dialect: ExportedProgram = export(Basic(), example_args, strict=True)

# Works correctly
print(aten_dialect.module()(torch.ones(3, 3), torch.ones(3, 3)))

# Errors
try:
    print(aten_dialect.module()(torch.ones(3, 2), torch.ones(3, 2)))
except Exception:
    tb.print_exc()

tensor([[2., 2., 2.],
        [2., 2., 2.],
        [2., 2., 2.]])
Traceback (most recent call last):
  File "/pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py", line 111, in <module>
    print(aten_dialect.module()(torch.ones(3, 2), torch.ones(3, 2)))
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/fx/graph_module.py", line 822, in call_wrapped
    return self._wrapped_call(self, *args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/fx/graph_module.py", line 400, in __call__
    raise e
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/fx/graph_module.py", line 387, in __call__
    return super(self.cls, obj).__call__(*args, **kwargs)  # type: ignore[misc]
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1739, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1845, in _call_impl
    return inner()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1772, in inner
    args_kwargs_result = hook(self, args, kwargs)  # type: ignore[misc]
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py", line 751, in _fn
    return fn(*args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/export/_unlift.py", line 49, in _check_input_constraints_pre_hook
    _check_input_constraints_for_graph(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/_export/utils.py", line 356, in _check_input_constraints_for_graph
    raise RuntimeError(
RuntimeError: Expected input at *args[0].shape[1] to be equal to 3, but got 2
To express that some input shapes are dynamic, we can insert dynamic

shapes to the exporting flow. This is done through the Dim API:

from torch.export import Dim


class Basic(torch.nn.Module):
    def __init__(self):
        super().__init__()

    def forward(self, x: torch.Tensor, y: torch.Tensor) -> torch.Tensor:
        return x + y


example_args = (torch.randn(3, 3), torch.randn(3, 3))
dim1_x = Dim("dim1_x", min=1, max=10)
dynamic_shapes = {"x": {1: dim1_x}, "y": {1: dim1_x}}
aten_dialect: ExportedProgram = export(
    Basic(), example_args, dynamic_shapes=dynamic_shapes, strict=True
)
print(aten_dialect)

ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, x: "f32[3, s0]", y: "f32[3, s0]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:127 in forward, code: return x + y
            add: "f32[3, s0]" = torch.ops.aten.add.Tensor(x, y);  x = y = None
            return (add,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='y'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='add'), target=None)])
Range constraints: {s0: VR[1, 10]}

Note that that the inputs arg0_1 and arg1_1 now have shapes (3, s0), with s0 being a symbol representing that this dimension can be a range of values.

Additionally, we can see in the Range constraints that value of s0 has the range [1, 10], which was specified by our dynamic shapes.

Now let’s try running the model with different shapes:

# Works correctly
print(aten_dialect.module()(torch.ones(3, 3), torch.ones(3, 3)))
print(aten_dialect.module()(torch.ones(3, 2), torch.ones(3, 2)))

# Errors because it violates our constraint that input 0, dim 1 <= 10
try:
    print(aten_dialect.module()(torch.ones(3, 15), torch.ones(3, 15)))
except Exception:
    tb.print_exc()

# Errors because it violates our constraint that input 0, dim 1 == input 1, dim 1
try:
    print(aten_dialect.module()(torch.ones(3, 3), torch.ones(3, 2)))
except Exception:
    tb.print_exc()

tensor([[2., 2., 2.],
        [2., 2., 2.],
        [2., 2., 2.]])
tensor([[2., 2.],
        [2., 2.],
        [2., 2.]])
Traceback (most recent call last):
  File "/pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py", line 154, in <module>
    print(aten_dialect.module()(torch.ones(3, 15), torch.ones(3, 15)))
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/fx/graph_module.py", line 822, in call_wrapped
    return self._wrapped_call(self, *args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/fx/graph_module.py", line 400, in __call__
    raise e
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/fx/graph_module.py", line 387, in __call__
    return super(self.cls, obj).__call__(*args, **kwargs)  # type: ignore[misc]
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1739, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1845, in _call_impl
    return inner()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1772, in inner
    args_kwargs_result = hook(self, args, kwargs)  # type: ignore[misc]
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py", line 751, in _fn
    return fn(*args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/export/_unlift.py", line 49, in _check_input_constraints_pre_hook
    _check_input_constraints_for_graph(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/_export/utils.py", line 344, in _check_input_constraints_for_graph
    raise RuntimeError(
RuntimeError: Expected input at *args[0].shape[1] to be <= 10, but got 15
Traceback (most recent call last):
  File "/pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py", line 160, in <module>
    print(aten_dialect.module()(torch.ones(3, 3), torch.ones(3, 2)))
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/fx/graph_module.py", line 822, in call_wrapped
    return self._wrapped_call(self, *args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/fx/graph_module.py", line 400, in __call__
    raise e
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/fx/graph_module.py", line 387, in __call__
    return super(self.cls, obj).__call__(*args, **kwargs)  # type: ignore[misc]
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1739, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1845, in _call_impl
    return inner()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1772, in inner
    args_kwargs_result = hook(self, args, kwargs)  # type: ignore[misc]
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py", line 751, in _fn
    return fn(*args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/export/_unlift.py", line 49, in _check_input_constraints_pre_hook
    _check_input_constraints_for_graph(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/_export/utils.py", line 308, in _check_input_constraints_for_graph
    raise RuntimeError(
RuntimeError: Expected input at *args[1].shape[1] to be equal to 3, but got 2

Addressing Untraceable Code

As our goal is to capture the entire computational graph from a PyTorch program, we might ultimately run into untraceable parts of programs. To address these issues, the torch.export documentation, or the torch.export tutorial would be the best place to look.

Performing Quantization

To quantize a model, we first need to capture the graph with torch.export.export_for_training, perform quantization, and then call torch.export. torch.export.export_for_training returns a graph which contains ATen operators which are Autograd safe, meaning they are safe for eager-mode training, which is needed for quantization. We will call the graph at this level, the Pre-Autograd ATen Dialect graph.

Compared to FX Graph Mode Quantization, we will need to call two new APIs: prepare_pt2e and convert_pt2e instead of prepare_fx and convert_fx. It differs in that prepare_pt2e takes a backend-specific Quantizer as an argument, which will annotate the nodes in the graph with information needed to quantize the model properly for a specific backend.

from torch.export import export_for_training

example_args = (torch.randn(1, 3, 256, 256),)
pre_autograd_aten_dialect = export_for_training(SimpleConv(), example_args).module()
print("Pre-Autograd ATen Dialect Graph")
print(pre_autograd_aten_dialect)

from torch.ao.quantization.quantize_pt2e import convert_pt2e, prepare_pt2e
from torch.ao.quantization.quantizer.xnnpack_quantizer import (
    get_symmetric_quantization_config,
    XNNPACKQuantizer,
)

quantizer = XNNPACKQuantizer().set_global(get_symmetric_quantization_config())
prepared_graph = prepare_pt2e(pre_autograd_aten_dialect, quantizer)
# calibrate with a sample dataset
converted_graph = convert_pt2e(prepared_graph)
print("Quantized Graph")
print(converted_graph)

aten_dialect: ExportedProgram = export(converted_graph, example_args, strict=True)
print("ATen Dialect Graph")
print(aten_dialect)

Pre-Autograd ATen Dialect Graph
GraphModule(
  (conv): Module()
)



def forward(self, x):
    x, = fx_pytree.tree_flatten_spec(([x], {}), self._in_spec)
    conv_weight = self.conv.weight
    conv_bias = self.conv.bias
    conv2d = torch.ops.aten.conv2d.default(x, conv_weight, conv_bias, [1, 1], [1, 1]);  x = conv_weight = conv_bias = None
    relu = torch.ops.aten.relu.default(conv2d);  conv2d = None
    return pytree.tree_unflatten((relu,), self._out_spec)

# To see more debug info, please use `graph_module.print_readable()`
/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/ao/quantization/utils.py:408: UserWarning: must run observer before calling calculate_qparams. Returning default values.
  warnings.warn(
/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/ao/quantization/observer.py:1318: UserWarning: must run observer before calling calculate_qparams.                                    Returning default scale and zero point
  warnings.warn(
Quantized Graph
GraphModule(
  (conv): Module()
)



def forward(self, x):
    x, = fx_pytree.tree_flatten_spec(([x], {}), self._in_spec)
    _frozen_param0 = self._frozen_param0
    dequantize_per_tensor_default = torch.ops.quantized_decomposed.dequantize_per_tensor.default(_frozen_param0, 1.0, 0, -127, 127, torch.int8);  _frozen_param0 = None
    conv_bias = self.conv.bias
    quantize_per_tensor_default_1 = torch.ops.quantized_decomposed.quantize_per_tensor.default(x, 1.0, 0, -128, 127, torch.int8);  x = None
    dequantize_per_tensor_default_1 = torch.ops.quantized_decomposed.dequantize_per_tensor.default(quantize_per_tensor_default_1, 1.0, 0, -128, 127, torch.int8);  quantize_per_tensor_default_1 = None
    conv2d = torch.ops.aten.conv2d.default(dequantize_per_tensor_default_1, dequantize_per_tensor_default, conv_bias, [1, 1], [1, 1]);  dequantize_per_tensor_default_1 = dequantize_per_tensor_default = conv_bias = None
    relu = torch.ops.aten.relu.default(conv2d);  conv2d = None
    quantize_per_tensor_default_2 = torch.ops.quantized_decomposed.quantize_per_tensor.default(relu, 1.0, 0, -128, 127, torch.int8);  relu = None
    dequantize_per_tensor_default_2 = torch.ops.quantized_decomposed.dequantize_per_tensor.default(quantize_per_tensor_default_2, 1.0, 0, -128, 127, torch.int8);  quantize_per_tensor_default_2 = None
    return pytree.tree_unflatten((dequantize_per_tensor_default_2,), self._out_spec)

# To see more debug info, please use `graph_module.print_readable()`
ATen Dialect Graph
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, p_conv_bias: "f32[16]", b__frozen_param0: "i8[16, 3, 3, 3]", x: "f32[1, 3, 256, 256]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:64 in forward, code: a = self.conv(x)
            dequantize_per_tensor: "f32[16, 3, 3, 3]" = torch.ops.quantized_decomposed.dequantize_per_tensor.default(b__frozen_param0, 1.0, 0, -127, 127, torch.int8);  b__frozen_param0 = None

             # File: <eval_with_key>.203:9 in forward, code: quantize_per_tensor_default_1 = torch.ops.quantized_decomposed.quantize_per_tensor.default(x, 1.0, 0, -128, 127, torch.int8);  x = None
            quantize_per_tensor: "i8[1, 3, 256, 256]" = torch.ops.quantized_decomposed.quantize_per_tensor.default(x, 1.0, 0, -128, 127, torch.int8);  x = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:64 in forward, code: a = self.conv(x)
            dequantize_per_tensor_1: "f32[1, 3, 256, 256]" = torch.ops.quantized_decomposed.dequantize_per_tensor.default(quantize_per_tensor, 1.0, 0, -128, 127, torch.int8);  quantize_per_tensor = None
            conv2d: "f32[1, 16, 256, 256]" = torch.ops.aten.conv2d.default(dequantize_per_tensor_1, dequantize_per_tensor, p_conv_bias, [1, 1], [1, 1]);  dequantize_per_tensor_1 = dequantize_per_tensor = p_conv_bias = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:65 in forward, code: return self.relu(a)
            relu: "f32[1, 16, 256, 256]" = torch.ops.aten.relu.default(conv2d);  conv2d = None
            quantize_per_tensor_1: "i8[1, 16, 256, 256]" = torch.ops.quantized_decomposed.quantize_per_tensor.default(relu, 1.0, 0, -128, 127, torch.int8);  relu = None

             # File: <eval_with_key>.203:14 in forward, code: dequantize_per_tensor_default_2 = torch.ops.quantized_decomposed.dequantize_per_tensor.default(quantize_per_tensor_default_2, 1.0, 0, -128, 127, torch.int8);  quantize_per_tensor_default_2 = None
            dequantize_per_tensor_2: "f32[1, 16, 256, 256]" = torch.ops.quantized_decomposed.dequantize_per_tensor.default(quantize_per_tensor_1, 1.0, 0, -128, 127, torch.int8);  quantize_per_tensor_1 = None
            return (dequantize_per_tensor_2,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.PARAMETER: 2>, arg=TensorArgument(name='p_conv_bias'), target='conv.bias', persistent=None), InputSpec(kind=<InputKind.BUFFER: 3>, arg=TensorArgument(name='b__frozen_param0'), target='_frozen_param0', persistent=True), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='dequantize_per_tensor_2'), target=None)])
Range constraints: {}

More information on how to quantize a model, and how a backend can implement a Quantizer can be found here.

Lowering to Edge Dialect

After exporting and lowering the graph to the ATen Dialect, the next step is to lower to the Edge Dialect, in which specializations that are useful for edge devices but not necessary for general (server) environments will be applied. Some of these specializations include:

  • DType specialization

  • Scalar to tensor conversion

  • Converting all ops to the executorch.exir.dialects.edge namespace.

Note that this dialect is still backend (or target) agnostic.

The lowering is done through the to_edge API.

from executorch.exir import EdgeProgramManager, to_edge

example_args = (torch.randn(1, 3, 256, 256),)
aten_dialect: ExportedProgram = export(SimpleConv(), example_args, strict=True)

edge_program: EdgeProgramManager = to_edge(aten_dialect)
print("Edge Dialect Graph")
print(edge_program.exported_program())

Edge Dialect Graph
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, p_conv_weight: "f32[16, 3, 3, 3]", p_conv_bias: "f32[16]", x: "f32[1, 3, 256, 256]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:64 in forward, code: a = self.conv(x)
            aten_convolution_default: "f32[1, 16, 256, 256]" = executorch_exir_dialects_edge__ops_aten_convolution_default(x, p_conv_weight, p_conv_bias, [1, 1], [1, 1], [1, 1], False, [0, 0], 1);  x = p_conv_weight = p_conv_bias = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:65 in forward, code: return self.relu(a)
            aten_relu_default: "f32[1, 16, 256, 256]" = executorch_exir_dialects_edge__ops_aten_relu_default(aten_convolution_default);  aten_convolution_default = None
            return (aten_relu_default,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.PARAMETER: 2>, arg=TensorArgument(name='p_conv_weight'), target='conv.weight', persistent=None), InputSpec(kind=<InputKind.PARAMETER: 2>, arg=TensorArgument(name='p_conv_bias'), target='conv.bias', persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_relu_default'), target=None)])
Range constraints: {}

to_edge() returns an EdgeProgramManager object, which contains the exported programs which will be placed on this device. This data structure allows users to export multiple programs and combine them into one binary. If there is only one program, it will by default be saved to the name “forward”.

class Encode(torch.nn.Module):
    def forward(self, x):
        return torch.nn.functional.linear(x, torch.randn(5, 10))


class Decode(torch.nn.Module):
    def forward(self, x):
        return torch.nn.functional.linear(x, torch.randn(10, 5))


encode_args = (torch.randn(1, 10),)
aten_encode: ExportedProgram = export(Encode(), encode_args, strict=True)

decode_args = (torch.randn(1, 5),)
aten_decode: ExportedProgram = export(Decode(), decode_args, strict=True)

edge_program: EdgeProgramManager = to_edge(
    {"encode": aten_encode, "decode": aten_decode}
)
for method in edge_program.methods:
    print(f"Edge Dialect graph of {method}")
    print(edge_program.exported_program(method))

Edge Dialect graph of encode
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, x: "f32[1, 10]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:261 in forward, code: return torch.nn.functional.linear(x, torch.randn(5, 10))
            aten_randn_default: "f32[5, 10]" = executorch_exir_dialects_edge__ops_aten_randn_default([5, 10], device = device(type='cpu'), pin_memory = False)
            aten_permute_copy_default: "f32[10, 5]" = executorch_exir_dialects_edge__ops_aten_permute_copy_default(aten_randn_default, [1, 0]);  aten_randn_default = None
            aten_mm_default: "f32[1, 5]" = executorch_exir_dialects_edge__ops_aten_mm_default(x, aten_permute_copy_default);  x = aten_permute_copy_default = None
            return (aten_mm_default,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_mm_default'), target=None)])
Range constraints: {}

Edge Dialect graph of decode
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, x: "f32[1, 5]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:266 in forward, code: return torch.nn.functional.linear(x, torch.randn(10, 5))
            aten_randn_default: "f32[10, 5]" = executorch_exir_dialects_edge__ops_aten_randn_default([10, 5], device = device(type='cpu'), pin_memory = False)
            aten_permute_copy_default: "f32[5, 10]" = executorch_exir_dialects_edge__ops_aten_permute_copy_default(aten_randn_default, [1, 0]);  aten_randn_default = None
            aten_mm_default: "f32[1, 10]" = executorch_exir_dialects_edge__ops_aten_mm_default(x, aten_permute_copy_default);  x = aten_permute_copy_default = None
            return (aten_mm_default,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_mm_default'), target=None)])
Range constraints: {}

We can also run additional passes on the exported program through the transform API. An in-depth documentation on how to write transformations can be found here.

Note that since the graph is now in the Edge Dialect, all passes must also result in a valid Edge Dialect graph (specifically one thing to point out is that the operators are now in the executorch.exir.dialects.edge namespace, rather than the torch.ops.aten namespace.

example_args = (torch.randn(1, 3, 256, 256),)
aten_dialect: ExportedProgram = export(SimpleConv(), example_args, strict=True)
edge_program: EdgeProgramManager = to_edge(aten_dialect)
print("Edge Dialect Graph")
print(edge_program.exported_program())

from executorch.exir.dialects._ops import ops as exir_ops
from executorch.exir.pass_base import ExportPass


class ConvertReluToSigmoid(ExportPass):
    def call_operator(self, op, args, kwargs, meta):
        if op == exir_ops.edge.aten.relu.default:
            return super().call_operator(
                exir_ops.edge.aten.sigmoid.default, args, kwargs, meta
            )
        else:
            return super().call_operator(op, args, kwargs, meta)


transformed_edge_program = edge_program.transform((ConvertReluToSigmoid(),))
print("Transformed Edge Dialect Graph")
print(transformed_edge_program.exported_program())

Edge Dialect Graph
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, p_conv_weight: "f32[16, 3, 3, 3]", p_conv_bias: "f32[16]", x: "f32[1, 3, 256, 256]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:64 in forward, code: a = self.conv(x)
            aten_convolution_default: "f32[1, 16, 256, 256]" = executorch_exir_dialects_edge__ops_aten_convolution_default(x, p_conv_weight, p_conv_bias, [1, 1], [1, 1], [1, 1], False, [0, 0], 1);  x = p_conv_weight = p_conv_bias = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:65 in forward, code: return self.relu(a)
            aten_relu_default: "f32[1, 16, 256, 256]" = executorch_exir_dialects_edge__ops_aten_relu_default(aten_convolution_default);  aten_convolution_default = None
            return (aten_relu_default,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.PARAMETER: 2>, arg=TensorArgument(name='p_conv_weight'), target='conv.weight', persistent=None), InputSpec(kind=<InputKind.PARAMETER: 2>, arg=TensorArgument(name='p_conv_bias'), target='conv.bias', persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_relu_default'), target=None)])
Range constraints: {}

Transformed Edge Dialect Graph
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, p_conv_weight: "f32[16, 3, 3, 3]", p_conv_bias: "f32[16]", x: "f32[1, 3, 256, 256]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:64 in forward, code: a = self.conv(x)
            aten_convolution_default: "f32[1, 16, 256, 256]" = executorch_exir_dialects_edge__ops_aten_convolution_default(x, p_conv_weight, p_conv_bias, [1, 1], [1, 1], [1, 1], False, [0, 0], 1);  x = p_conv_weight = p_conv_bias = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:65 in forward, code: return self.relu(a)
            aten_sigmoid_default: "f32[1, 16, 256, 256]" = executorch_exir_dialects_edge__ops_aten_sigmoid_default(aten_convolution_default);  aten_convolution_default = None
            return (aten_sigmoid_default,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.PARAMETER: 2>, arg=TensorArgument(name='p_conv_weight'), target='conv.weight', persistent=None), InputSpec(kind=<InputKind.PARAMETER: 2>, arg=TensorArgument(name='p_conv_bias'), target='conv.bias', persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_sigmoid_default'), target=None)])
Range constraints: {}

Note: if you see error like torch._export.verifier.SpecViolationError: Operator torch._ops.aten._native_batch_norm_legit_functional.default is not Aten Canonical, please file an issue in https://github.com/pytorch/executorch/issues and we’re happy to help!

Delegating to a Backend

We can now delegate parts of the graph or the whole graph to a third-party backend through the to_backend API. An in-depth documentation on the specifics of backend delegation, including how to delegate to a backend and how to implement a backend, can be found here.

There are three ways for using this API:

  1. We can lower the whole module.

  2. We can take the lowered module, and insert it in another larger module.

  3. We can partition the module into subgraphs that are lowerable, and then lower those subgraphs to a backend.

Lowering the Whole Module

To lower an entire module, we can pass to_backend the backend name, the module to be lowered, and a list of compile specs to help the backend with the lowering process.

class LowerableModule(torch.nn.Module):
    def __init__(self):
        super().__init__()

    def forward(self, x):
        return torch.sin(x)


# Export and lower the module to Edge Dialect
example_args = (torch.ones(1),)
aten_dialect: ExportedProgram = export(LowerableModule(), example_args, strict=True)
edge_program: EdgeProgramManager = to_edge(aten_dialect)
to_be_lowered_module = edge_program.exported_program()

from executorch.exir.backend.backend_api import LoweredBackendModule, to_backend

# Import the backend
from executorch.exir.backend.test.backend_with_compiler_demo import (  # noqa
    BackendWithCompilerDemo,
)

# Lower the module
lowered_module: LoweredBackendModule = to_backend(
    "BackendWithCompilerDemo", to_be_lowered_module, []
)
print(lowered_module)
print(lowered_module.backend_id)
print(lowered_module.processed_bytes)
print(lowered_module.original_module)

# Serialize and save it to a file
save_path = "delegate.pte"
with open(save_path, "wb") as f:
    f.write(lowered_module.buffer())

LoweredBackendModule()
BackendWithCompilerDemo
b'1version:0#op:demo::aten.sin.default, numel:1, dtype:torch.float32<debug_handle>2#'
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, x: "f32[1]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:355 in forward, code: return torch.sin(x)
            aten_sin_default: "f32[1]" = executorch_exir_dialects_edge__ops_aten_sin_default(x);  x = None
            return (aten_sin_default,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_sin_default'), target=None)])
Range constraints: {}

In this call, to_backend will return a LoweredBackendModule. Some important attributes of the LoweredBackendModule are:

  • backend_id: The name of the backend this lowered module will run on in the runtime

  • processed_bytes: a binary blob which will tell the backend how to run this program in the runtime

  • original_module: the original exported module

Compose the Lowered Module into Another Module

In cases where we want to reuse this lowered module in multiple programs, we can compose this lowered module with another module.

class NotLowerableModule(torch.nn.Module):
    def __init__(self, bias):
        super().__init__()
        self.bias = bias

    def forward(self, a, b):
        return torch.add(torch.add(a, b), self.bias)


class ComposedModule(torch.nn.Module):
    def __init__(self):
        super().__init__()
        self.non_lowerable = NotLowerableModule(torch.ones(1) * 0.3)
        self.lowerable = lowered_module

    def forward(self, x):
        a = self.lowerable(x)
        b = self.lowerable(a)
        ret = self.non_lowerable(a, b)
        return a, b, ret


example_args = (torch.ones(1),)
aten_dialect: ExportedProgram = export(ComposedModule(), example_args, strict=True)
edge_program: EdgeProgramManager = to_edge(aten_dialect)
exported_program = edge_program.exported_program()
print("Edge Dialect graph")
print(exported_program)
print("Lowered Module within the graph")
print(exported_program.graph_module.lowered_module_0.backend_id)
print(exported_program.graph_module.lowered_module_0.processed_bytes)
print(exported_program.graph_module.lowered_module_0.original_module)

/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/export/_unlift.py:75: UserWarning: Attempted to insert a get_attr Node with no underlying reference in the owning GraphModule! Call GraphModule.add_submodule to add the necessary submodule, GraphModule.add_parameter to add the necessary Parameter, or nn.Module.register_buffer to add the necessary buffer
  getattr_node = gm.graph.get_attr(lifted_node)
/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/fx/graph.py:1801: UserWarning: Node non_lowerable_bias target non_lowerable.bias bias of non_lowerable does not reference an nn.Module, nn.Parameter, or buffer, which is what 'get_attr' Nodes typically target
  warnings.warn(
Edge Dialect graph
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, c_non_lowerable_bias: "f32[1]", x: "f32[1]"):
             # File: /opt/conda/envs/py_3.10/lib/python3.10/site-packages/executorch/exir/lowered_backend_module.py:344 in forward, code: return executorch_call_delegate(self, *args)
            lowered_module_0 = self.lowered_module_0
            executorch_call_delegate: "f32[1]" = torch.ops.higher_order.executorch_call_delegate(lowered_module_0, x);  lowered_module_0 = x = None

             # File: /opt/conda/envs/py_3.10/lib/python3.10/site-packages/executorch/exir/lowered_backend_module.py:344 in forward, code: return executorch_call_delegate(self, *args)
            lowered_module_1 = self.lowered_module_0
            executorch_call_delegate_1: "f32[1]" = torch.ops.higher_order.executorch_call_delegate(lowered_module_1, executorch_call_delegate);  lowered_module_1 = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:409 in forward, code: return torch.add(torch.add(a, b), self.bias)
            aten_add_tensor: "f32[1]" = executorch_exir_dialects_edge__ops_aten_add_Tensor(executorch_call_delegate, executorch_call_delegate_1)
            aten_add_tensor_1: "f32[1]" = executorch_exir_dialects_edge__ops_aten_add_Tensor(aten_add_tensor, c_non_lowerable_bias);  aten_add_tensor = c_non_lowerable_bias = None
            return (executorch_call_delegate, executorch_call_delegate_1, aten_add_tensor_1)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.CONSTANT_TENSOR: 4>, arg=TensorArgument(name='c_non_lowerable_bias'), target='non_lowerable.bias', persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='executorch_call_delegate'), target=None), OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='executorch_call_delegate_1'), target=None), OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_add_tensor_1'), target=None)])
Range constraints: {}

Lowered Module within the graph
BackendWithCompilerDemo
b'1version:0#op:demo::aten.sin.default, numel:1, dtype:torch.float32<debug_handle>2#'
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, x: "f32[1]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:355 in forward, code: return torch.sin(x)
            aten_sin_default: "f32[1]" = executorch_exir_dialects_edge__ops_aten_sin_default(x);  x = None
            return (aten_sin_default,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_sin_default'), target=None)])
Range constraints: {}

Notice that there is now a torch.ops.higher_order.executorch_call_delegate node in the graph, which is calling lowered_module_0. Additionally, the contents of lowered_module_0 are the same as the lowered_module we created previously.

Partition and Lower Parts of a Module

A separate lowering flow is to pass to_backend the module that we want to lower, and a backend-specific partitioner. to_backend will use the backend-specific partitioner to tag nodes in the module which are lowerable, partition those nodes into subgraphs, and then create a LoweredBackendModule for each of those subgraphs.

class Foo(torch.nn.Module):
    def forward(self, a, x, b):
        y = torch.mm(a, x)
        z = y + b
        a = z - a
        y = torch.mm(a, x)
        z = y + b
        return z


example_args = (torch.randn(2, 2), torch.randn(2, 2), torch.randn(2, 2))
aten_dialect: ExportedProgram = export(Foo(), example_args, strict=True)
edge_program: EdgeProgramManager = to_edge(aten_dialect)
exported_program = edge_program.exported_program()
print("Edge Dialect graph")
print(exported_program)

from executorch.exir.backend.test.op_partitioner_demo import AddMulPartitionerDemo

delegated_program = to_backend(exported_program, AddMulPartitionerDemo())
print("Delegated program")
print(delegated_program)
print(delegated_program.graph_module.lowered_module_0.original_module)
print(delegated_program.graph_module.lowered_module_1.original_module)

Edge Dialect graph
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, a: "f32[2, 2]", x: "f32[2, 2]", b: "f32[2, 2]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:455 in forward, code: y = torch.mm(a, x)
            aten_mm_default: "f32[2, 2]" = executorch_exir_dialects_edge__ops_aten_mm_default(a, x)

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:456 in forward, code: z = y + b
            aten_add_tensor: "f32[2, 2]" = executorch_exir_dialects_edge__ops_aten_add_Tensor(aten_mm_default, b);  aten_mm_default = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:457 in forward, code: a = z - a
            aten_sub_tensor: "f32[2, 2]" = executorch_exir_dialects_edge__ops_aten_sub_Tensor(aten_add_tensor, a);  aten_add_tensor = a = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:458 in forward, code: y = torch.mm(a, x)
            aten_mm_default_1: "f32[2, 2]" = executorch_exir_dialects_edge__ops_aten_mm_default(aten_sub_tensor, x);  aten_sub_tensor = x = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:459 in forward, code: z = y + b
            aten_add_tensor_1: "f32[2, 2]" = executorch_exir_dialects_edge__ops_aten_add_Tensor(aten_mm_default_1, b);  aten_mm_default_1 = b = None
            return (aten_add_tensor_1,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='a'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='b'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_add_tensor_1'), target=None)])
Range constraints: {}

Delegated program
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, a: "f32[2, 2]", x: "f32[2, 2]", b: "f32[2, 2]"):
            # No stacktrace found for following nodes
            lowered_module_0 = self.lowered_module_0
            lowered_module_1 = self.lowered_module_1
            executorch_call_delegate_1 = torch.ops.higher_order.executorch_call_delegate(lowered_module_1, a, x, b);  lowered_module_1 = None
            getitem_1: "f32[2, 2]" = executorch_call_delegate_1[0];  executorch_call_delegate_1 = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:457 in forward, code: a = z - a
            aten_sub_tensor: "f32[2, 2]" = executorch_exir_dialects_edge__ops_aten_sub_Tensor(getitem_1, a);  getitem_1 = a = None

            # No stacktrace found for following nodes
            executorch_call_delegate = torch.ops.higher_order.executorch_call_delegate(lowered_module_0, aten_sub_tensor, x, b);  lowered_module_0 = aten_sub_tensor = x = b = None
            getitem: "f32[2, 2]" = executorch_call_delegate[0];  executorch_call_delegate = None
            return (getitem,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='a'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='b'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='getitem'), target=None)])
Range constraints: {}

ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, aten_sub_tensor: "f32[2, 2]", x: "f32[2, 2]", b: "f32[2, 2]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:458 in forward, code: y = torch.mm(a, x)
            aten_mm_default_1: "f32[2, 2]" = executorch_exir_dialects_edge__ops_aten_mm_default(aten_sub_tensor, x);  aten_sub_tensor = x = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:459 in forward, code: z = y + b
            aten_add_tensor_1: "f32[2, 2]" = executorch_exir_dialects_edge__ops_aten_add_Tensor(aten_mm_default_1, b);  aten_mm_default_1 = b = None
            return [aten_add_tensor_1]

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='aten_sub_tensor'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='b'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_add_tensor_1'), target=None)])
Range constraints: {}

ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, a: "f32[2, 2]", x: "f32[2, 2]", b: "f32[2, 2]"):
             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:455 in forward, code: y = torch.mm(a, x)
            aten_mm_default: "f32[2, 2]" = executorch_exir_dialects_edge__ops_aten_mm_default(a, x);  a = x = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:456 in forward, code: z = y + b
            aten_add_tensor: "f32[2, 2]" = executorch_exir_dialects_edge__ops_aten_add_Tensor(aten_mm_default, b);  aten_mm_default = b = None
            return [aten_add_tensor]

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='a'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='b'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_add_tensor'), target=None)])
Range constraints: {}

Notice that there are now 2 torch.ops.higher_order.executorch_call_delegate nodes in the graph, one containing the operations add, mul and the other containing the operations mul, add.

Alternatively, a more cohesive API to lower parts of a module is to directly call to_backend on it:

class Foo(torch.nn.Module):
    def forward(self, a, x, b):
        y = torch.mm(a, x)
        z = y + b
        a = z - a
        y = torch.mm(a, x)
        z = y + b
        return z


example_args = (torch.randn(2, 2), torch.randn(2, 2), torch.randn(2, 2))
aten_dialect: ExportedProgram = export(Foo(), example_args, strict=True)
edge_program: EdgeProgramManager = to_edge(aten_dialect)
exported_program = edge_program.exported_program()
delegated_program = edge_program.to_backend(AddMulPartitionerDemo())

print("Delegated program")
print(delegated_program.exported_program())

Delegated program
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, a: "f32[2, 2]", x: "f32[2, 2]", b: "f32[2, 2]"):
            # No stacktrace found for following nodes
            lowered_module_0 = self.lowered_module_0
            lowered_module_1 = self.lowered_module_1
            executorch_call_delegate_1 = torch.ops.higher_order.executorch_call_delegate(lowered_module_1, a, x, b);  lowered_module_1 = None
            getitem_1: "f32[2, 2]" = executorch_call_delegate_1[0];  executorch_call_delegate_1 = None

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:491 in forward, code: a = z - a
            aten_sub_tensor: "f32[2, 2]" = executorch_exir_dialects_edge__ops_aten_sub_Tensor(getitem_1, a);  getitem_1 = a = None

            # No stacktrace found for following nodes
            executorch_call_delegate = torch.ops.higher_order.executorch_call_delegate(lowered_module_0, aten_sub_tensor, x, b);  lowered_module_0 = aten_sub_tensor = x = b = None
            getitem: "f32[2, 2]" = executorch_call_delegate[0];  executorch_call_delegate = None
            return (getitem,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='a'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='b'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='getitem'), target=None)])
Range constraints: {}

Running User-Defined Passes and Memory Planning

As a final step of lowering, we can use the to_executorch() API to pass in backend-specific passes, such as replacing sets of operators with a custom backend operator, and a memory planning pass, to tell the runtime how to allocate memory ahead of time when running the program.

A default memory planning pass is provided, but we can also choose a backend-specific memory planning pass if it exists. More information on writing a custom memory planning pass can be found here

from executorch.exir import ExecutorchBackendConfig, ExecutorchProgramManager
from executorch.exir.passes import MemoryPlanningPass

executorch_program: ExecutorchProgramManager = edge_program.to_executorch(
    ExecutorchBackendConfig(
        passes=[],  # User-defined passes
        memory_planning_pass=MemoryPlanningPass(),  # Default memory planning pass
    )
)

print("ExecuTorch Dialect")
print(executorch_program.exported_program())

import executorch.exir as exir

ExecuTorch Dialect
ExportedProgram:
    class GraphModule(torch.nn.Module):
        def forward(self, a: "f32[2, 2]", x: "f32[2, 2]", b: "f32[2, 2]"):
            # No stacktrace found for following nodes
            alloc: "f32[2, 2]" = executorch_exir_memory_alloc(((2, 2), torch.float32))

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:489 in forward, code: y = torch.mm(a, x)
            aten_mm_default: "f32[2, 2]" = torch.ops.aten.mm.out(a, x, out = alloc);  alloc = None

            # No stacktrace found for following nodes
            alloc_1: "f32[2, 2]" = executorch_exir_memory_alloc(((2, 2), torch.float32))

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:490 in forward, code: z = y + b
            aten_add_tensor: "f32[2, 2]" = torch.ops.aten.add.out(aten_mm_default, b, out = alloc_1);  aten_mm_default = alloc_1 = None

            # No stacktrace found for following nodes
            alloc_2: "f32[2, 2]" = executorch_exir_memory_alloc(((2, 2), torch.float32))

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:491 in forward, code: a = z - a
            aten_sub_tensor: "f32[2, 2]" = torch.ops.aten.sub.out(aten_add_tensor, a, out = alloc_2);  aten_add_tensor = a = alloc_2 = None

            # No stacktrace found for following nodes
            alloc_3: "f32[2, 2]" = executorch_exir_memory_alloc(((2, 2), torch.float32))

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:492 in forward, code: y = torch.mm(a, x)
            aten_mm_default_1: "f32[2, 2]" = torch.ops.aten.mm.out(aten_sub_tensor, x, out = alloc_3);  aten_sub_tensor = x = alloc_3 = None

            # No stacktrace found for following nodes
            alloc_4: "f32[2, 2]" = executorch_exir_memory_alloc(((2, 2), torch.float32))

             # File: /pytorch/executorch/docs/source/tutorials_source/export-to-executorch-tutorial.py:493 in forward, code: z = y + b
            aten_add_tensor_1: "f32[2, 2]" = torch.ops.aten.add.out(aten_mm_default_1, b, out = alloc_4);  aten_mm_default_1 = b = alloc_4 = None
            return (aten_add_tensor_1,)

Graph signature: ExportGraphSignature(input_specs=[InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='a'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='x'), target=None, persistent=None), InputSpec(kind=<InputKind.USER_INPUT: 1>, arg=TensorArgument(name='b'), target=None, persistent=None)], output_specs=[OutputSpec(kind=<OutputKind.USER_OUTPUT: 1>, arg=TensorArgument(name='aten_add_tensor_1'), target=None)])
Range constraints: {}

Notice that in the graph we now see operators like torch.ops.aten.sub.out and torch.ops.aten.div.out rather than torch.ops.aten.sub.Tensor and torch.ops.aten.div.Tensor.

This is because between running the backend passes and memory planning passes, to prepare the graph for memory planning, an out-variant pass is run on the graph to convert all of the operators to their out variants. Instead of allocating returned tensors in the kernel implementations, an operator’s out variant will take in a prealloacated tensor to its out kwarg, and store the result there, making it easier for memory planners to do tensor lifetime analysis.

We also insert alloc nodes into the graph containing calls to a special executorch.exir.memory.alloc operator. This tells us how much memory is needed to allocate each tensor output by the out-variant operator.

Saving to a File

Finally, we can save the ExecuTorch Program to a file and load it to a device to be run.

Here is an example for an entire end-to-end workflow:

import torch
from torch.export import export, export_for_training, ExportedProgram


class M(torch.nn.Module):
    def __init__(self):
        super().__init__()
        self.param = torch.nn.Parameter(torch.rand(3, 4))
        self.linear = torch.nn.Linear(4, 5)

    def forward(self, x):
        return self.linear(x + self.param).clamp(min=0.0, max=1.0)


example_args = (torch.randn(3, 4),)
pre_autograd_aten_dialect = export_for_training(M(), example_args).module()
# Optionally do quantization:
# pre_autograd_aten_dialect = convert_pt2e(prepare_pt2e(pre_autograd_aten_dialect, CustomBackendQuantizer))
aten_dialect: ExportedProgram = export(
    pre_autograd_aten_dialect, example_args, strict=True
)
edge_program: exir.EdgeProgramManager = exir.to_edge(aten_dialect)
# Optionally do delegation:
# edge_program = edge_program.to_backend(CustomBackendPartitioner)
executorch_program: exir.ExecutorchProgramManager = edge_program.to_executorch(
    ExecutorchBackendConfig(
        passes=[],  # User-defined passes
    )
)

with open("model.pte", "wb") as file:
    file.write(executorch_program.buffer)

Conclusion

In this tutorial, we went over the APIs and steps required to lower a PyTorch program to a file that can be run on the ExecuTorch runtime.

Docs

Access comprehensive developer documentation for PyTorch

View Docs

Tutorials

Get in-depth tutorials for beginners and advanced developers

View Tutorials

Resources

Find development resources and get your questions answered

View Resources

To analyze traffic and optimize your experience, we serve cookies on this site. By clicking or navigating, you agree to allow our usage of cookies. As the current maintainers of this site, Facebook’s Cookies Policy applies. Learn more, including about available controls: Cookies Policy.