Kannan Ramchandran

Sample Complexity and Representation Ability of Test-time Scaling Paradigms

SPECS: Faster Test-Time Scaling through Speculative Drafts and Dynamic Switching

Proxy-SPEX: Sample-Efficient Interpretability via Sparse Feature Interactions in LLMs

SPEX: Scaling Feature Interaction Explanations for LLMs

EmbedLLM: Learning Compact Representations of Large Language Models

VersaPRM: Multi-Domain Process Reward Model via Synthetic Reasoning Data

Looped Transformers for Length Generalization

Watermarking using Semantic-aware Speculative Sampling: from Theory to Practice

Bias Learning: Quantifying and Mitigating Position Sensitivity in Text Embeddings

Beyond Cosine Similarity: Introducing the Unified semantic Similarity Metric Benchmark (USMB) for Text Similarity Measurement