Executorch | Not Hacker News!

Discussion (15 comments)

Showing 22 comments

13 days ago

2 replies

I've heard from a friend who works in the embedded space that Tensorflow Lite is still the only realistic (supported by vendors) game in town for running ML models on microcontrollers such as ESP32, nRF, etc. The hardware support listed for this project seems like it's targeting much "fatter" MCUs (Android, etc).

nickpsecurity

13 days ago

1 reply

Yeah, it's targeting "micro"-controllers, not microcontrollers. I was hoping for a PyTorch solution to TF Lite.

This is still great, though. Previously, I thought a mobile model (eg speech/object recognition) would require me to learn both PyTorch and something like MLC in C++. Then, port them.

If this is as it appears, I could develop a small model that could run on mobile on my laptop, train it on cloud GPU's, test it locally, and use this tool to produce a mobile version (or save some steps?). That would keep us from having to learn C++ or MLC just to do mobile.

I mean, one still can learn other tools for their advantages. However, ML students and startups might benefit greatly from this by being able to rapidly develop or port mobile apps. Then, people learning other tools for their advantages build stuff that way. The overall ecosystem gets stronger with more competition.

orignldrgibl

13 days ago

1 reply

I'll plug: https://github.com/google-ai-edge/ai-edge-torch for torch to tflite conversion.

nickpsecurity

13 days ago

I was hoping something like that existed, too. Thanks for the link!

lukeinator42

13 days ago

yeah that checks out, although looks like they do have an example for running models on a raspberry pi pico 2: https://docs.pytorch.org/executorch/main/pico2_tutorial.html. The list of embedded platforms this can run on is probably greater than the list of backends, it just wouldn't have acceleration.

fooblaster

13 days ago

2 replies

I am so confused by metas ecosystem. Perhaps others have the same issues. I have mountains of torchscript code. It worked fine for me - had no issues making the python compatible. Torchscript is now deprecated, and the ostensible replacement is torch.export and either: AOTInductor or executorch. torch.export is so limited - no control flow at runtime at all, less support of python than torchscript. It is far more work to hoist all the control flow out of the model than it ever was to make the model torchscript compatible. Feel like meta has moved on, but I'm still stuck in the past here.

ThouYS

13 days ago

1 reply

it's quite the bummer. some models you simply can't export with dynamo. for the time being the jit exporter is the only good option.

in particular selective function scripting is essential!

drag0nblad3

13 days ago

3 replies

ExecuTorch developer here, agreed it's a huge pain to deal with if conditions right now. Part of the pain comes from the vast expressiveness of python on if condition, which causes all ML compiler a lot of headache to be able to capture a sound graph. The rest of the pain comes from the strict requirement of torch.compile itself (no mutation/aliasing behavior in the if branches), which in often times makes torch.cond hard to use or inefficient.

fooblaster

10 days ago

So what are your users doing to get around this? Hoisting all control flow out?

fooblaster

9 days ago

Anyway, perhaps we can chat in the executorch discord.

fooblaster

9 days ago

And you wouldn't happen to know about a torchscript replacement that is currently in-flight that is not based on export?

sorrow17

13 days ago

1 reply

Yeah, for a lot of users who control the exported source code, rewriting model to use control flow ops, or simply removing the control flow code is a viable option and solvable. For some other users who want to export the model as-is, the option is either using the (deprecated) torchscript, or just move on and use torch.compile and run your model in Python.

fooblaster

13 days ago

Those control flow ops aren't even supported on many backends. I know tensor rt doesn't support them for example, at least today.

Removing control flow isn't as easy as you'd think for some. It essentially means ripping large sections out of python and into separately implemented c++.

lewisjoe

13 days ago

2 replies

It'd be great if it supports a wasm/web backend as well.

I bet a lot of trivial text capabilities (grammar checking, autocomplete, etc) will benefit from this rather than sending everything to a hosted model. It's possible right now with onnx / transformers.js / tensorflow.js - but isn't quite there yet in terms of efficiency. Given the target for microcontrollers, it'd be great to bring that efficiency to browsers as well.

klaussilveira

13 days ago

If you need WASM, I think Candle is your current best bet: https://github.com/huggingface/candle

miguel_martin

13 days ago

You can compile to wasm, I have done so via the XNNPACK backend - you might have to tweak the compilation settings and upgrade the XNNPACK submodule/patch some code. But this only supports CPU, not a WebGPU or WebGL backend.

r2vcap

13 days ago

1 reply

I get the impression that https://github.com/pytorch/executorch is Meta’s take on TFLite / LiteRT, which is quite interesting.

While reading the README and related documentation, I noticed that Samsung Exynos NPU acceleration was listed, which immediately caught my attention. According to https://docs.pytorch.org/executorch/main/backends/samsung/sa..., Samsung has finally built and released an NPU SDK—so I followed the link to check it out.

Unfortunately, the experience was disappointing.

The so-called “version 1.0” SDK is available only for Ubuntu 22.04 / 20.04. There is no release date information per version, nor any visible roadmap. Even worse, downloading the SDK requires logging in. The product description page itself https://soc-developer.semiconductor.samsung.com/global/devel... does contain explanations, but they are provided almost entirely as images rather than text—presented in a style more reminiscent of corporate PR material than developer-facing technical documentation.

This is, regrettably, very typical of Samsung’s software support: opaque documentation, gated access, and little consideration for external developers. At this point, it is hard not to conclude that Exynos remains a poor choice, regardless of its theoretical hardware capabilities.

For comparison, Qualcomm and MediaTek actively collaborate with existing ecosystems, and their SDKs are generally available without artificial barriers. As a concrete example, see how LiteRT distributes its artifacts and references in this commit: https://github.com/google-ai-edge/LiteRT/commit/eaf7d635e1bc...

my123

12 days ago

Is https://github.com/Samsung/ENNDelegate enough or is it TFLite/LiteRT only?

aero88883

13 days ago

My observation working with Tensorflow Lite was that the framework overheads are almost 15-20% of the total processing cycles. Hope executorch is a lighter framework.

executorch

12 days ago

ExecuTorch member here.

- Better microcontroller support is in our roadmap for 2026. There is a lot of development happening here from support for Arduino, STMicro and others. We will do this openly with the community as usual so if you are interested, feel free to join our discord and looped into the github repo.

- Better web support is also in the roadmap. There is some limited support already though not sure exactly what your usecase is. Feel free to open up a GH issue and we can see if there is a way to unblock you.

- Will take the feedback about Samsung to them. Seeing the user feedback first hand here will likely help them prioritize some of that. This is partially why we have not called this a production ready backend unlike the other backends like Qualcomm, Vulkan and a few others we ourselves are using in production.

captaindiego

12 days ago

How does performance stack up against TensorRT for edge NVidia hardware?

stuaxo

13 days ago

So the vulkan backend for pytorch is just in executorch?

I just want it on native desktop python.

Resources