HKUST

Activity Feed Request to join this org

AI & ML interests

None defined yet.

Recent Activity

hubin submitted a paper 4 days ago

CorrGRPO: Correlation-Normalized GRPO for Multi-Reward Learning

TalentZHOU submitted a paper about 2 months ago

Hierarchical Self-Improvement: A Framework for Task-Specific Evolvable Agent Harnesses

Yushi98 submitted a paper 2 months ago

Fewer Clarifications, Better Code: Benchmarking Cross-Session Personalized Ambiguity Adaptation in Coding Assistants

View all activity

Papers

CorrGRPO: Correlation-Normalized GRPO for Multi-Reward Learning

Hierarchical Self-Improvement: A Framework for Task-Specific Evolvable Agent Harnesses

View all Papers

HKUST 's papers 24

Submitted by

Wenbin Hu

CorrGRPO: Correlation-Normalized GRPO for Multi-Reward Learning

Submitted by

Tailin

Hierarchical Self-Improvement: A Framework for Task-Specific Evolvable Agent Harnesses

Submitted by

Yushi SUN

The Personalization Mirage: How LLMs Fabricate User Profiles, and Why Self-Monitoring Misleads

3

Submitted by

Yushi SUN

Fewer Clarifications, Better Code: Benchmarking Cross-Session Personalized Ambiguity Adaptation in Coding Assistants

3

CMI-Mem: Toward Generalizable Long-Term Memory Management via CMI-Augmented Reinforcement Learning

Submitted by

Yushi SUN

Navigating the Mirage: A Dual-Path Agentic Framework for Robust Misleading Chart Question Answering

2

Submitted by

Yushi SUN

Are LLMs Ready for Scientific Discovery? A Capability-Oriented Benchmark for AI Scientists

2

Submitted by

zzhongyj

AsySplat: Efficient Asymmetric 3D Gaussian Splatting for Long-Sequence Scene Modeling

Submitted by

Lu Dai

Towards Mechanistically Understanding Why Memorized Knowledge Fails to Generalize in Large Language Model Finetuning

Submitted by

Geng Xinyu

DeepSearch-World: Self-Distillation for Deep Search Agents in a Verifiable Environment

Submitted by

Tianlin Pan

NormGuard: Reward-Preserving Norm Constraints in Flow-Matching Reinforcement Learning

2

Submitted by

Wang Yanghao

LISA: Likelihood Score Alignment for Visual-condition Controllable Generation