Raphael Pisoni ยท X (formerly Twitter)

user avatar

I love science, technology and engineering, I'm fascinated by the speed and power of innovation and i can't wait for the future! ML Researcher

Austria

Joined March 2013

  • Pinned

    user avatar

    Neural networks have a fundamental problem. Feed them garbage data and instead of admitting that they are confused, they will confidently hallucinate. I just open-sourced the HALO-Loss to try and fix this. It give the model a mathematically sound *I don't know!* button.๐Ÿงต

  • user avatar

    What if I told you that Scaled Dot-Product Attention (SDPA) doesn't make a whole lot of sense for ViTs and you can get faster convergence, better performance and near-linear scaling with a few well-steered Gaussians? Let me introduce you to SSOG (Separable Sum Of Gaussians)!

  • user avatar

    Will be working on the weekend! New blogpost coming soon!๐Ÿ˜‰

  • user avatar

    Insane time to be alive...

  • user avatar

    I mean...

Read the original on x.com โ†—