Yash Kant

I am a Research Visitor at Georgia Tech supervised by Devi Parikh, Dhruv Batra, Peter Anderson, and Harsh Agrawal. Presently, I am working on the intersection of Computer Vision and Natural Language Processing.

I finished my undergraduate studies from Indian Institute of Technology Roorkee. I have interned at Microsoft, Bangalore and visited National University of Singapore twice as a research assistant.

Email  /  CV  /  Github  /  Twitter  /  LinkedIn  /  Instagram  /  Facebook  / 

profile photo

  • Adding Complement Objective Training to Pythia: I experimented with adding Complement Objective Training in FAIR's vision and language framework Pythia and also wrote a report on my findings here, the code is here.
  • ICLR Reproducibility Challenge: We reproduced Closing the Generalization Gap of Adaptive Gradient Methods in Training Deep Neural Networks and here's the code.
  • Visual Chatbot Version 2.0 (here): I shifted the old Lua-Torch codebase to PyTorch, added better captioning and trained the VisDial model on BUTD features.
  • Quantized Neural Architecture Search (unreleased): I quantized the search space of Neural Architecture Search algorithms [ENAS and PNAS] to search for resource-efficient models.

I borrowed this template from Jon Barron's website,