diff --git a/docs/source/backends-xnnpack.md b/docs/source/backends-xnnpack.md index 77285d3cdbe..b07543f5c68 100644 --- a/docs/source/backends-xnnpack.md +++ b/docs/source/backends-xnnpack.md @@ -81,6 +81,10 @@ quantizer.set_global(quantization_config) #### Quantizing a model with the XNNPACKQuantizer After configuring the quantizer, the model can be quantized by via the `prepare_pt2e` and `convert_pt2e` APIs. ```python +from torch.ao.quantization.quantize_pt2e import ( + prepare_pt2e, + convert_pt2e, +) from torch.export import export_for_training exported_model = export_for_training(model_to_quantize, example_inputs).module() diff --git a/docs/source/getting-started.md b/docs/source/getting-started.md index 492334c81df..4d8e3f0189d 100644 --- a/docs/source/getting-started.md +++ b/docs/source/getting-started.md @@ -5,26 +5,25 @@ This section is intended to describe the necessary steps to take PyTorch model a - Run the model using the ExecuTorch runtime APIs on your development platform. - Deploy the model to the target platform using the ExecuTorch runtime. +## System Requirements +The following are required to install the ExecuTorch host libraries, needed to export models and run from Python. Requirements for target end-user devices are backend dependent. See the appropriate backend documentation for more information. + +- Python 3.10 - 3.12 +- g++ version 7 or higher, clang++ version 5 or higher, or another C++17-compatible toolchain. +- Linux or MacOS operating system (Arm or x86). + - Windows is supported via WSL. + ## Installation -To use ExecuTorch, you will need to install both the Python package and the appropriate platform-specific runtime libraries. +To use ExecuTorch, you will need to install both the Python package and the appropriate platform-specific runtime libraries. Pip is the recommended way to install the ExecuTorch python package. -Pip is the recommended way to install the ExecuTorch python package. This package includes the dependencies needed to export a PyTorch model, as well as Python runtime bindings for model testing and evaluation. It is common to install the package within a Python virtual environment, in order to meet the Python and dependency version requirements. +This package includes the dependencies needed to export a PyTorch model, as well as Python runtime bindings for model testing and evaluation. Consider installing ExecuTorch within a virtual environment, such as one provided by [conda](https://docs.conda.io/projects/conda/en/latest/user-guide/getting-started.html#creating-environments) or [venv](https://packaging.python.org/en/latest/guides/installing-using-pip-and-virtual-environments/#create-and-use-virtual-environments). ``` pip install executorch ``` -To build the framework from source, see [Building From Source](using-executorch-building-from-source.md). - -Backend delegates may require additional dependencies. See the appropriate backend documentation for more information. +To build the framework from source, see [Building From Source](using-executorch-building-from-source.md). Backend delegates may require additional dependencies. See the appropriate backend documentation for more information. -#### System Requirements -The following are required to install the ExecuTorch host libraries, needed to export models and run from Python. Requirements for target end-user devices are backend dependent. See the appropriate backend documentation for more information. - -- Python 3.10 - 3.12 -- g++ version 7 or higher, clang++ version 5 or higher, or another C++17-compatible toolchain. -- Linux or MacOS operating system (Arm or x86). - - Windows is supported via WSL.
@@ -44,15 +43,20 @@ ExecuTorch provides hardware acceleration for a wide variety of hardware. The mo For mobile use cases, consider using XNNPACK for Android and Core ML or XNNPACK for iOS as a first step. See [Hardware Backends](backends-overview.md) for more information. ### Exporting -Exporting is done using Python APIs. ExecuTorch provides a high degree of customization during the export process, but the typical flow is as follows: +Exporting is done using Python APIs. ExecuTorch provides a high degree of customization during the export process, but the typical flow is as follows. This example uses the MobileNet V2 image classification model implementation in torchvision, but the process supports any [export-compliant](https://pytorch.org/docs/stable/export.html) PyTorch model. + ```python -import executorch +import torch +import torchvision.models as models +from torchvision.models.mobilenetv2 import MobileNet_V2_Weights +from executorch.backends.xnnpack.partition.xnnpack_partitioner import XnnpackPartitioner +from executorch.exir import to_edge_transform_and_lower -model = MyModel() # The PyTorch model to export -example_inputs = (torch.randn(1,3,64,64),) # A tuple of inputs +model = models.mobilenetv2.mobilenet_v2(weights=MobileNet_V2_Weights.DEFAULT).eval() +sample_inputs = (torch.randn(1, 3, 224, 224), ) -et_program = executorch.exir.to_edge_transform_and_lower( - torch.export.export(model, example_inputs), +et_program = to_edge_transform_and_lower( + torch.export.export(model, sample_inputs), partitioner=[XnnpackPartitioner()] ).to_executorch() @@ -62,7 +66,7 @@ with open("model.pte", "wb") as f: If the model requires varying input sizes, you will need to specify the varying dimensions and bounds as part of the `export` call. See [Model Export and Lowering](using-executorch-export.md) for more information. -The hardware backend to target is controlled by the partitioner parameter to to\_edge\_transform\_and\_lower. In this example, the XnnpackPartitioner is used to target mobile CPUs. See the delegate-specific documentation for a full description of the partitioner and available options. +The hardware backend to target is controlled by the partitioner parameter to to\_edge\_transform\_and\_lower. In this example, the XnnpackPartitioner is used to target mobile CPUs. See the [backend-specific documentation](backends-overview.md) for information on how to use each backend. Quantization can also be done at this stage to reduce model size and runtime. Quantization is backend-specific. See the documentation for the target backend for a full description of supported quantization schemes. @@ -70,16 +74,19 @@ Quantization can also be done at this stage to reduce model size and runtime. Qu After successfully generating a .pte file, it is common to use the Python runtime APIs to validate the model on the development platform. This can be used to evaluate model accuracy before running on-device. -Inference can be run as follows: +For the MobileNet V2 model from torchvision used in this example, image inputs are expected as a normalized, float32 tensor with a dimensions of (batch, channels, height, width). The output See [torchvision.models.mobilenet_v2](https://pytorch.org/vision/main/models/generated/torchvision.models.mobilenet_v2.html) for more information on the input and output tensor format for this model. + ```python +import torch from executorch.runtime import Runtime +from typing import List runtime = Runtime.get() -input_tensor = torch.randn(1,3,128,128) -program = runtime.load_program("/path/to/mode.pte") +input_tensor: torch.Tensor = torch.randn(1, 3, 224, 224) +program = runtime.load_program("model.pte") method = program.load_method("forward") -outputs = method.execute([input_tensor]) +outputs: List[torch.Tensor] = method.execute([input_tensor]) ``` @@ -101,13 +108,15 @@ To add the library to your app, download the AAR, and add it to the gradle build ``` mkdir -p app/libs -curl https://ossci-android.s3.amazonaws.com/executorch/release/executorch-241002/executorch.aar -o app/libs/executorch.aar +curl https://ossci-android.s3.amazonaws.com/executorch/release/v0.5.0-rc3/executorch.aar -o app/libs/executorch.aar ``` And in gradle, ``` # app/build.gradle.kts dependencies { implementation(files("libs/executorch.aar")) + implementation("com.facebook.soloader:soloader:0.10.5") + implementation("com.facebook.fbjni:fbjni:0.5.1") } ``` diff --git a/docs/source/using-executorch-export.md b/docs/source/using-executorch-export.md index 9bcfedc256a..fb51ff1bd40 100644 --- a/docs/source/using-executorch-export.md +++ b/docs/source/using-executorch-export.md @@ -59,7 +59,7 @@ class Model(torch.nn.Module): torch.nn.Conv2d(8, 16, 3), torch.nn.ReLU(), torch.nn.AdaptiveAvgPool2d((1,1)) - ) + ) self.linear = torch.nn.Linear(16, 10) def forward(self, x): @@ -68,12 +68,14 @@ class Model(torch.nn.Module): y = self.linear(y) return y -model = Model() +model = Model().eval() inputs = (torch.randn(1,1,16,16),) outputs = model(*inputs) print(f"Model output: {outputs}") ``` +Note that the model is set to evaluation mode using `.eval()`. Models should always be exported in evaluation mode unless performing on-device training. This mode configures certain operations with training-specific behavior, such as batch norm or dropout, to use the inference-mode configuration. + ## Export and Lowering To actually export and lower the model, call `export`, `to_edge_transform_and_lower`, and `to_executorch` in sequence. This yields an ExecuTorch program which can be serialized to a file. Putting it all together, lowering the example model above using the XNNPACK delegate for mobile CPU performance can be done as follows: @@ -92,7 +94,7 @@ class Model(torch.nn.Module): torch.nn.ReLU(), torch.nn.Conv2d(8, 16, 3), torch.nn.ReLU(), - torch.nn.AdaptiveAvgPool2d([1,1]) + torch.nn.AdaptiveAvgPool2d((1,1)) ) self.linear = torch.nn.Linear(16, 10)