Ke Li

Attend to the Active: Structure-Aware Dynamic Attention in LLMs for Compositional Instruction Following

Count Counts: Motivating Exploration in LLM Reasoning with Count-based Intrinsic Rewards

FlexibleLLM: Making Low-Bit Quantization for Large Language Models More Flexible and Efficient

DeepOmni: Towards Seamless and Smart Speech Interaction with Adaptive Modality-Specific MoE

CUARewardBench: Benchmark for Evaluating Reward Models on Computer-using Agent Trajectories

HGT-UCOD: A Hint-Guided Teacher Framework for Unsupervised Camouflaged Object Detection

Breaking the Bias: Quantifying the Attention of Industrial Anomaly Detection

Learn the Ropes, Then Trust the Wins: Self-imitation with Progressive Exploration for Agentic Reinforcement Learning

Training-Free Group Relative Policy Optimization

RoRecomp: Enhancing Reasoning Efficiency via Rollout Response Recomposition in Reinforcement Learning

Pseudo-Label Supervision in Unsupervised Industrial Anomaly Detection

VITA-1.5: Towards GPT-4o Level Real-Time Vision and Speech Interaction

Zooming from Context to Cue: Hierarchical Preference Optimization for Multi-Image MLLMs

VITA-Audio: Fast Interleaved Audio-Text Token Generation for Efficient Large Speech-Language Model

Incentivizing Reasoning for Advanced Instruction-Following of Large Language Models

FlowAgent: a New Paradigm for Workflow Agent

Detecting and Evaluating Medical Hallucinations in Large Vision Language Models

Less is More: Learning Reference Knowledge Using No-Reference Image Quality Assessment

Freeze-Omni: A Smart and Low Latency Speech-to-speech Dialogue Model with Frozen LLM