Towards Memory-Efficient Training for Extremely Large Output Spaces : Learning with 670k Labels on a Single Commodity GPU

Hakutulokset

• Aktiivinen

  ScaleX/Babbar: Scalable and Robust Representation Learning in Large output Spaces

  Babbar, R. (Vastuullinen tutkija)

  01/09/2022 → 31/08/2026

  Projekti: RCF Academy Project

  • Multi-Label Classification 100%
  • Classification Problem 69%
  • Performance Metric 45%
  • Text Classification 44%
  • Graphics Processing Unit 39%
• Päättynyt

  HPC-HD/Babbar: High Performance Computing for the Detection and Analysis of Historical Discourses

  Babbar, R. (Vastuullinen tutkija)

  01/01/2022 → 31/12/2024

  Projekti: RCF Academy Project targeted call

  • Multi-Label Classification 100%
  • Classification Problem 60%
  • Performance Metric 39%
  • Text Classification 38%
  • Graphics Processing Unit 34%