COLM 2026

856 papers

Top Research Directions

Research directions most represented by COLM 2026 papers.

Activation Steering for Inference Time Model Control 21
The Causal Disconnect in Language Model Reasoning 19
Shifting from Static Benchmarks to Dynamic Evaluation 18
Uncertainty Calibration for Large Language Models 18
Scaling Language Model Reasoning Beyond Verifiable Rewards 16
Mitigating Computational Overthinking in Large Reasoning Models 15
General Purpose Reasoning from Specialized Reinforcement Learning 14
Group Relative Policy Optimization for Large Reasoning Models 14
Activation Patching for Mechanistic Interpretability 13
Autonomous Code Refinement via Execution Feedback Loops 13
Iterative Retrieval Augmented Generation for Multi Hop Reasoning 12
Evaluating Generative AI with Large Language Models 12
The Erosion of Safety Guardrails in Multi-Turn Dialogue 10
Role Specialization in Multi-Agent Architectures 10
Understanding Prompt Sensitivity in Language Models 10

Top Authors

AuthorPapers
Dakai Guo1
Junqi Liu1
Junqi Liu1
Lihong Zhi1
Ruichen Qiu1
Ruichen Qiu1
Ruyong Feng1
Xiao-Shan Gao1
Xiao-Shan Gao1
Yichuan Cao1

Papers

Title
Why Your Prompt Compressor Is Deleting the Wrong Tokens: An Information-Theoretic Diagnosis
Wiener Filtering for VLM Hallucination Suppression
WildTableBench: Benchmarking Multimodal Foundation Models on Table Understanding In the Wild
Wired for Overconfidence: A Mechanistic Perspective on Inflated Verbalized Confidence in LLMs
WorldPM: Scaling Human Preference Modeling via Real-World Feedback
YC-Bench: Benchmarking AI Agents for Long-Term Planning and Consistent Execution.

« prev Page 18/18