Inderjit S Dhillon

The Art of Scaling Reinforcement Learning Compute for LLMs

LLM-guided Hierarchical Retrieval

LUCID: Attention with Preconditioned Representations

SLIM: Structure-aware Low-rank Inference Model

LoRA Done RITE: Robust Invariant Transformation Equilibration for LoRA Optimization

Retraining with Predicted Hard Labels Provably Increases Model Accuracy

Large Language Models are Interpretable Learners

Scalable In-context Ranking with Generative Models

LASER: Attention using Exponential Transformation

Geometric Median (GM) Matching for Robust k-Subset Selection from Noisy Data

LASER: Attention with Exponential Transformation

Geometric Median (GM) Matching for Robust Data Pruning

Retraining with Predicted Hard Labels Provably Increases Model Accuracy