Dimitris Papailiopoulos

Sample More to Think Less: Group Filtered Policy Optimization for Concise Reasoning

Not All Bits Are Equal: Scale-Dependent Memory Optimization Strategies for Reasoning Models

ReJump: A Tree-Jump Representation for Analyzing and Improving LLM Reasoning

Let's Let's Let's Let's... Understand Looping in Reasoning Models

Compress to Think, Decompress to Speak: Dual-Mode Reasoning in Transformers

Extrapolation by Association: Length Generalization Transfer In Transformers

Everything Everywhere All at Once: LLMs can In-Context Learn Multiple Tasks in Superposition

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

Self-Improving Transformers Overcome Easy-to-Hard and Length Generalization Challenges

Task Vectors in In-Context Learning: Emergence, Formation, and Benefits

From Artificial Needles to Real Haystacks: Improving Retrieval Capabilities in LLMs by Finetuning on Synthetic Data

Everything Everywhere All at Once: LLMs can In-Context Learn Multiple Tasks in Superposition

Lexico: Extreme KV Cache Compression via Sparse Coding over Universal Dictionaries

KV-Dict: Sparse KV Cache Compression with Universal Dictionaries