Seed2.1 Model Card: Agentic Intelligence for Productivity
A model family focused on dependable agent workflows, software engineering, and practical productivity.
Team model cardPaper & resources
AI RESEARCHER · BYTEDANCE SEED
Qinyu Luo.
How can agents become
thoughtful collaborators?
I’m currently at ByteDance Seed. I study language agents that reason, use tools, and work together. My interests span multi-agent systems, agentic tool learning, and multimodal reasoning—with an eye toward systems that are useful beyond the benchmark.
ACADEMIC BACKGROUND
THUNLP, Tsinghua UniversityRESEARCH EXPERIENCE
Johns Hopkins UniversityGRADUATE STUDIES
A little more about me 01 / PUBLICATIONS & REPORTS
Research, in the open.
The complete collection: 14 works across agent systems, model learning, and mathematics. Key figures from the papers accompany each work.
14 works10 per page
Seed2.0 Model Card: Towards Intelligence Frontier for Real-World Complexity
A model family combining reasoning, multimodal understanding, and agentic execution for complex real-world tasks.
Team model cardPaper & resources
Rethinking On-Policy Distillation of Large Language Models II: One Training Example
Investigating why on-policy distillation can recover much of full-data training’s gain from a single query.
PreprintPaper & resources
Two Boundary-Certificate Proofs of the Crouzeix Theorem: Fiberwise Fourier Recurrence and Defect-Gram Telescoping
A preprint proposing two candidate proof routes for the finite-dimensional scalar Crouzeix inequality.
Preprint · not peer reviewedPaper & resources
Test-Time Deep Thinking to Explore Implicit Rules
A framework that helps agents infer hidden environmental rules by separating action execution from deeper exploration.
KDD 2026Paper & resources
Current Agents Fail to Leverage World Model as Tool for Foresight
An empirical study of whether vision-language agents can use external world models to anticipate action outcomes.
ACL 2026Paper & resources
Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence
A self-evolving training arena that synthesizes executable environments and verifiable tasks from real-world themes.
PreprintPaper & resources
Seed1.8 Model Card: Towards Generalized Real-World Agency
A team model card describing a foundation model for multi-turn interaction, tool use, and complex execution.
Team model cardPaper & resources
UI-TARS-2 Technical Report: Advancing GUI Agent with Multi-Turn Reinforcement Learning
A native GUI agent trained for multi-turn interaction across desktop, mobile, and game environments.
Technical reportPaper & resources
EscapeBench: Towards Advancing Creative Intelligence of Language Model Agents
Room-escape environments for studying creative problem-solving, implicit goals, and unexpected uses of familiar tools.
Conference paperPaper & resources
Distance between Relevant Information Pieces Causes Bias in Long-Context LLMs
Studying how the spacing between related facts changes a language model’s ability to use a long context.
Conference paperPaper & resources
Proactive Agent: Shifting LLM Agents from Reactive Responses to Active Assistance
Learning when an agent should offer help, using observations of user activity and feedback on proposed tasks.
Conference paperPaper & resources
RepoAgent: An LLM-Powered Open-Source Framework for Repository-level Code Documentation Generation
An open-source agent that generates and maintains documentation with a view of the whole code repository.
Conference paperPaper & resources
XAgent: An Autonomous Agent for Complex Task Solving
An open-source autonomous agent system for planning and carrying out complex tasks with external tools.
Team project reportPaper & resources
No papers found.
Try another keyword or research area.