Xin Zhao

WSM: Decay-Free Learning Rate Schedule via Checkpoint Merging for LLM Pre-training

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

Intermediate Layers Can Be Self-Hard Negative Generator For Large Language Model Based Recommendation

Unleashing Perception-Time Scaling to Multimodal Reasoning Models

Latent Reasoning with Recurrent Depth for Sequential Recommendation

Towards High Data Efficiency in Reinforcement Learning with Verifiable Reward

ReForm: Reflective Autoformalization with Prospective Bounded Sequence Optimization

Enhancing Cross-task Transfer of Large Language Models via Fourier Activation Steering

ForesightKV: Optimizing KV Cache Eviction for Reasoning Models by Learning Long-Term Contribution

PAL-UI: Planning with Active Look-back for Vision-Based GUI Agents

IterResearch: Rethinking Long-Horizon Agents with Interaction Scaling

Pass@k Training for Adaptively Balancing Exploration and Exploitation of Large Reasoning Models

MARS: Optimizing Dual-System Deep Research via Multi-Agent Reinforcement Learning

Experience-Guided Reflective Co-Evolution of Prompts and Heuristics for Automatic Algorithm Design

Irrational Complex Rotations Empower Low-bit Optimizers

Incentivizing Dual Process Thinking for Efficient Large Language Model Reasoning

Neuron based Personality Trait Induction in Large Language Models

Exploring the Design Space of Visual Context Representation in Video MLLMs

Investigating the Pre-Training Dynamics of In-Context Learning: Task Recognition vs. Task Learning

C-3PO: Compact Plug-and-Play Proxy Optimization to Achieve Human-like Retrieval-Augmented Generation

Domain-Specific Pruning of Large Mixture-of-Experts Models with Few-shot Demonstrations

LESS IS MORE: HIGH-VALUE DATA SELECTION FOR VISUAL INSTRUCTION TUNING

Extracting and Transferring Abilities For Building Multi-lingual Ability-enhanced Large Language Models

Mix-CPT: A Domain Adaptation Framework via Decoupling Knowledge Learning and Format Alignment

MOEfication by Experts as Masks