Cunxiang Wang

TrustJudge: Inconsistencies of LLM-as-a-Judge and How to Alleviate Them

Deep Literature Survey Automation with an Iterative Workflow

Rethinking LLM Evaluation: Can We Evaluate LLMs with 200× Less Data?

AudioMarathon: A Comprehensive Benchmark for Long-Context Audio Understanding and Efficiency in Audio LLMs

RLAR: An Agentic Reward System for Multi-task Reinforcement Learning on Large Language Models

Detecting Variant Contamination in LLMs via Variance of Generation Distribution

PROBE: Benchmarking Reasoning Paradigm Overfitting in Large Language Models

SPaR: Self-Play with Tree-Search Refinement to Improve Instruction-Following in Large Language Models

NovelQA: Benchmarking Question Answering on Documents Exceeding 200K Tokens

RedHat: Towards Reducing Hallucination in Essay Critiques with Large Language Models