Suhas Kotha

Hello! I'm a senior studying CS at CMU, advised by Aditi Raghunathan. I am currently researching how to reliably adapt pretrained models for downstream tasks. In the past, I was a quantitative trading intern at Jane Street and a research intern at NASA SpaceML.

Papers

Writing

A collection of random experiments, thoughts, and introductions.

• Challenge: Language Data Ordering/Filtering
• Synthesizing Optimal Parallel Collectives via SAT
• Challenge: Emergence as Prediction Across Scale
• Elegance of Linear Regression
• Paper Analysis: Provable Guarantees for Self-Supervised Deep Learning with Spectral Contrastive Loss
• Understanding Reinforcement Learning from Human Feedback
• Convex Geometry for Computers
• The Type Theory of Transformers
• Paper Analysis: A Universal Law of Robustness via Isoperimetry