stardance has been extended another month! the new deadline is october 31 :)

You are browsing as a guest. Sign up (or log in) to start making projects!

KANVAS

By @Swapnil_The_Curious

  • 1 Devlogs
  • 3 Total hours

KANVAS is a model that predicts whether a job on a shared GPU cluster will fail, and shows the reason behind every prediction. The curves add up to a failure probability, so any prediction breaks down into ten pieces which could be checked. It is being trained on 18,890 tasks from Alibaba’s trace and it matched a black box, 0.752 AUC against 0.756.

Source
Ship #1 Pending review

@Swapnil_The_Curious on KANVAS

I made KANVAS, a failure predictor for GPU cluster jobs built to be auditable. I wanted to show a readable model can keep up with a black box on real data. The hardest part was the data. My first telemetry join kept only 2 failures in 996 rows, because the trace records telemetry mostly for jobs that succeed, so I swapped in neighbouring jobs’ telemetry. I’m proudest that I used the model to attack itself. Its best signal was one user’s habit, and on unseen users all three models fall to chance.

  • 1 devlog
  • 3h
Try project → See source code →

Delete project?

Are you sure you want to permanently delete this project? This action cannot be undone.

All devlogs, followers, and associated data will be removed.

Followers

Loading…