| layout | title | permalink | background-class | body-class | order | published |
|---|---|---|---|---|---|---|
get_started |
PyTorch for Edge |
/get-started/executorch/ |
get-started-background |
get-started |
5 |
true |
PyTorch’s edge specific library is [ExecuTorch(https://github.com/pytorch/executorch/)] and is designed to be lightweight, very performant even on devices with constrained hardware such as mobile phones, embedded systems and microcontrollers.
ExecuTorch relies heavily on PyTorch core technologies such as torch.compile and torch.export, and should be very familiar with anyone who has used PyTorch in the past.
You can get started by following the general getting started guide or jump to the specific steps for your target device.
Using ExecuTorch on Android Using ExecuTorch on iOS Using ExecuTorch with C++
ExecuTorch provides out of the box hardware acceleration for a growing number of chip manufacturers. See the following resources to learn more about how to leverage them.
Backend Overview XNNPACK Core ML MPS Vulkan ARM Ethos-U Qualcomm AI Engine MediaTek Cadence Xtensa
<script page-id="mobile" src="{{ site.baseurl }}/assets/menu-tab-selection.js"></script> <script src="{{ site.baseurl }}/assets/get-started-sidebar.js"></script>