[Formerly the Metis M.2.] Maximise your Axelera Embedded devices – connect, ask, and share to optimise your AI projects with this powerful edge accelerator
Recently active
Hello,I know that axelera does not work in the attention mechanism. For example, can we convert the attention mechanism in the yolov11 architecture to convolution? Have you done any research or read on this subject? Does Softmax structure work in Axelera? (as far as I know, no) Is there a way to turn Softmax into a working structure?
Hello, is there a detailed document where I can learn about the Axelera chip architecture and how to do the quantization in it?Thanks
Hello,I wanted to test the Yolo11n model. But I couldn't do this because I got an error in the calibrating part. Is this because of the use of attention mechanism in Yolo11 and Yolo12?Error: ValueError: Some modules couldn’t be simplified: <… activation.Softmax> <…. matmul.MatMul>In addition, when I looked at the ready yaml files on voyager sdk github, I saw that the yolov8x model was not available. Is there a special reason for this? Because when I prepared and tried a new yolov8x yaml file, I got the similar error again. But as far as I know, the attention mechanism is not used in Yolov8x.Thanks
I’m scoping out the Metis M.2 module for a compact edge AI project and wondering how capable it really is. I’m mainly interested in real-time object detection on 1080p video streams, maybe 2–4 cameras max.Can it handle multi-stream inputs in real-time, or is that better suited to the PCIe version?
It’s actually not that complicated getting started with an M.2 module, and there’s a good chance that the Metis isn’t the first M.2 you’ve ever had in your hands.But it’s still good policy to familiarise yourself with both the hardware and the software when you’re ready to unbox that little AI beauty.To help with that, we’ve got a series of articles to walk you through…Installation: Handling and installing the M.2 in your host system, and a great tip on how to confirm there’s suitable power getting to it once it’s in place Integration: Top tips on figuring out whether the host PC (or system or platform) is nicely compatible with the needs of the Metis accelerator. Voyager SDK: Once it’s installed and ready, all you need is the Voyager SDK.This is all just to get you started, and then the real work begins! Make sure you share what you’re building right here on the Axelera community 👍
voyager // wingman
Describe your vision pipeline in plain words. Wingman writes the YAML, runs it live on your Metis, and irons out the bugs with you.
the prompt // now open
Pitch it for a chance to win one of 10 Dell XE5s + Metis PCIe to build it on. Closes soon.
Already have an account? Login
No account yet? Create an account
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.