You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
An established workflow for deploying ML models to the zync exists and an excellent video has been made about exactly this in the context of large experimental physics control systems by Josh at RadiaSoft https://www.youtube.com/watch?v=NWpcP_sFRm4. I met him at ICALEPCS this year and he explained that many of their filtering algorithms are being replaced with ML autoencoders in this manner. Further, im not sure if its interesting but in the same workflow it is actually possible to deploy any arbitrary algorithm on the Zync DPU written in Python PyTorch/numpy (it does not have to be something that is trained).
Is it an idea to add an ML model panda block?
The text was updated successfully, but these errors were encountered:
An established workflow for deploying ML models to the zync exists and an excellent video has been made about exactly this in the context of large experimental physics control systems by Josh at RadiaSoft https://www.youtube.com/watch?v=NWpcP_sFRm4. I met him at ICALEPCS this year and he explained that many of their filtering algorithms are being replaced with ML autoencoders in this manner. Further, im not sure if its interesting but in the same workflow it is actually possible to deploy any arbitrary algorithm on the Zync DPU written in Python PyTorch/numpy (it does not have to be something that is trained).
Is it an idea to add an ML model panda block?
The text was updated successfully, but these errors were encountered: