Shijin Wang

Fewer Battles, More Gain: An Information-Efficient Framework for Arena-based LLM Evaluation

Exploring Expert Concentration for Parameter-efficient Fine-tuning of Mixture-of-Expert LLMs

ChemEval: A Multi-level and Fine-grained Chemical Capability Evaluation for Large Language Models

Beneath the Surface: Exposing and Mitigating Surface Learning in Large Language Models

From Values to Tokens: An LLM-Driven Framework for Context-aware Time Series Forecasting via Symbolic Discretization

FACT: Mitigating Inconsistent Hallucinations in LLMs via Fact-Driven Alternating Code-Text Training

am-ELO: A Stable Framework for Arena-based LLM Evaluation

How Does Sequence Modeling Architecture Influence Base Capabilities of Pre-trained Language Models? Exploring Key Architecture Design Principles to Avoid Base Capabilities Degradation

CogMath: Assessing LLMs' Authentic Mathematical Ability from a Human Cognitive Perspective

Investigating and Mitigating Catastrophic Forgetting in Medical Knowledge Injection through Internal Knowledge Augmentation Learning

A Closed-Form Solution for Fast and Reliable Adaptive Testing

CursorCore: Assist Programming through Aligning Anything

Evaluating LLMs Across Multi-Cognitive Levels: From Medical Knowledge Mastery to Scenario-Based Problem Solving

CursorCore: Assist Programming through Aligning Anything

Evaluating Large Language Models through Role-Guide and Self-Reflection: A Comparative Study

CogMath: Evaluating LLMs' Authentic Mathematical Ability from a Cognitive Perspective