Qipeng Guo

JanusCoder: Towards a Foundational Visual-Programmatic Interface for Code Intelligence

Beyond Real: Imaginary Extension of Rotary Position Embeddings for Long-Context LLMs

Context-level Language Modeling by Learning Predictive Context Embeddings

MLP Memory: A Retriever-Pretrained Memory for Large Language Models

Explicit Multi-head Attention for Inter-head Interaction in Large Language Models

Code-driven Number Sequence Calculation: Enhancing the Inductive Reasoning Abilities of Large Language Models

IFDECORATOR: Wrapping Instruction Following Reinforcement Learning with Verifiable Rewards

GeometryZero: Advancing LLM Geometry Solving via Group Contrastive Policy Optimization

Lost in Real-World Scenarios: Concretization Disrupts LLM Logical Reasoning

Beyond Homogeneous Attention: Memory-Efficient LLMs via Fourier-Approximated KV Cache

EvoSyn: Generalizable Evolutionary Data Synthesis for Verifiable Learning

InteractScience: Programmatic and Visually-Grounded Evaluation of Interactive Scientific Demonstration Code Generation

Windtalkers: Watermarking Open-Source LLMs with Ciphered-Instruction

VideoRoPE: What Makes for Good Video Rotary Position Embedding?

Memory Decoder: A Pretrained, Plug-and-Play Memory for Large Language Models

Pre-Trained Policy Discriminators are General Reward Models

World-aware Planning Narratives Enhance Large Vision-Language Model Planner

NovelQA: Benchmarking Question Answering on Documents Exceeding 200K Tokens

Implicit Reward as the Bridge: A Unified View of SFT and DPO Connections

ReAttention: Training-Free Infinite Context with Finite Attention Scope

Mixing Expert Knowledge: Bring Human Thoughts Back To the Game of Go

What are the Essential Factors in Crafting Effective Long Context Multi-Hop Instruction Datasets? Insights and Best Practices

BALCONI: BALancing CONtext and Internal Knowledge For Training Flexible LLMs

GAOKAO-Eval: Does High Scores Truly Reflect Strong Capabilities in LLMs?

LongSafetyBench: Long-Context LLMs Struggle with Safety Issues