Junyao Yang

About Me

Hi there, this is Junyao. I am a first-year graduate student at the School of Computing, National University of Singapore (NUS), specializing in Artificial Intelligence, and currently a Research Intern at Tencent Hy Frontier Lab, Singapore. My current research interests focus on Agentic AI (Agentic RL, Recursive Self-Improving, Long-Horizon Terminus Agent), Reinforcement Learning (RL Stability), and Trustworthy Artificial Intelligence. My Chinese name is 杨竣尧 (/jɑːŋ dʒuːn jaʊ/).
CVs: EN, 中文

News

2026.10 We release Terminal-RSI! Recursively coupling verified trajectory rewriting with critic-inheriting policy refinement to improve terminal agents. Project Page.
2026.09 Tech report: T1. A terminal agent for long-horizon tasks, featuring stable reinforcement learning training and dense verification rewards.
2026.08 Conducting my Dissertation under the guidance of Professor Shuicheng Yan.
2026.04 Joined Tencent Hy LLM Team, working on Agentic RL Stability and Self-Evolving Long-Horizon Terminus Agent.
2026.04 First-Author paper ReasonAny accepted to ACL 2026 Main.
2026.01 Tech report: AgentDoG. State-of-the-art diagnostic guardrail framework with an Agentic XAI attribution module.
2025.11 First-Author paper RCP-Merging accepted to AAAI 2026 Main Track.
2025.08 RewardDS accepted to EMNLP 2025 Main.
2025.08 Joined Shanghai AI Lab as a Research Intern, working with Dongrui Liu.
2025.05 Co-First-Author paper PrivacyRestore accepted to ACL 2025 Main.
2024.07 Joined ZeroNLP as a Research Assistant, advised by Prof. Ziqian Zeng.
2024.07 Contextless CS reached 20,000 DAU.
2024.04 Joined Tencent as a machine learning intern.
2024.03 Completed internship at ShenZhen Stock Exchange as a machine learning intern.

Selected Publications

arXiv Preprint Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine

Junyao Yang*, Zongxia Li*, Yucheng Shi*, Zhongzhi Li, Ruhan Wang, Ruiwen Zhou, Haitao Mi, Shuicheng Yan^†, Leowei Liang

TL;DR: Alternating verified trajectory rewriting with critic-inheriting policy refinement, using turn-boundary credit assignment and stage-specific rewards to advance terminal agents from partial progress to strict task completion; Terminal-RSI-27B achieves 19.84% on Terminal-Bench 4.0 and 86.52% on Terminal-Bench 2.1.

Paper Project Page

Tech Report (Core Contributor)

T1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks

Junyao Yang*^†, Yucheng Shi*, Zhongzhi Li, Ruhan Wang, Zongxia Li, Haitao Mi, Leowei Liang

TL;DR: A 122B MoE terminal agent trained by reinforcement learning — with stable MoE RL training framework and dense verification reward design aiming for Long-Horizon terminal tasks.

Paper Project Page Model

ACL 2026 ReasonAny: Incorporating Reasoning Capability to Any Model via Simple and Effective Model Merging

Junyao Yang, Chen Qian, Dongrui Liu^†, Wen Shen, Yong Liu^†, Jing Shao^†

TL;DR: Merging robust chain-of-thought capabilities into domain-specific models (Safety, Biomedicine) using Contrastive Gradient Identification.

Paper Code

AAAI 2026 RCP-Merging: Merging Long Chain-of-Thought Models with Domain-Specific Models by Considering Reasoning Capability as Prior

Junyao Yang, Jianwei Wang, Huiping Zhuang, Cen Chen, Ziqian Zeng*^†

TL;DR: Enhancing domain performance while preserving chain-of-thought reasoning abilities by treating reasoning as a prior.

Paper Code

ACL 2025 PrivacyRestore: Privacy-Preserving Inference in Large Language Models via Privacy Removal and Restoration

Ziqian Zeng*^†, Jianwei Wang*, Junyao Yang*, Zhengdong Lu, Haoran Li, Huiping Zhuang, Cen Chen

TL;DR: Protecting privacy via activation steering using a protected meta-vector without retraining.

Paper Code

Tech Report

AgentDoG: A Diagnostic Guardrail Framework for AI Agent Safety and Security

Shanghai Artificial Intelligence Laboratory (Contributor)

TL;DR: A state-of-the-art diagnostic guardrail framework utilizing a unified three-dimensional taxonomy to provide fine-grained monitoring and root-cause analysis of AI agent safety risks.

Paper Code Model #1 Paper of the day 机器之心

arXiv Preprint RSR: Scaling Trajectories for Complex Tasks through Recursive Self-Rewrite

Zongxia Li^†, Yucheng Shi^†, Zhongzhi Li^†, Junyao Yang, Ruhan Wang, Chengsong Huang, Fuxiao Liu, Haitao Mi, Jordan Boyd-Graber, Leowei Liang

TL;DR: Recursively rewriting successful trajectories discovered under diverse harnesses into reusable capabilities under a general harness, expanding 2,001 to 11,094 high-quality trajectories and lifting Qwen-3.8-27B pass@3 from 57.0% to 74.2% on Terminal-Bench 2.

Paper Model

arXiv Preprint SWR: Self-Supervised Scaling of Terminal Environments for Scientific Domains

Zhongzhi Li*, Yucheng Shi*, Zongxia Li*, Junyao Yang, Ruhan Wang, Yu Wang, Jingyuan Huang, Jichao Yu, Ninghao Liu^†, Haitao Mi, Leowei Liang

TL;DR: Software-in-the-loop reconstruction derives reference behavior and verifiers from existing scientific software workflows, scaling to 500 workflows across 46 software families and six domains and improving Qwen3.8-27B on Terminal-Bench 2 from 47.94% to 53.56%.

Paper

arXiv Preprint Long-Horizon-Terminal-Bench: Testing the Limits of Agents on Long-Horizon Terminal Tasks with Dense Reward-Based Grading

Zongxia Li, Zhongzhi Li, Yucheng Shi, Ruhan Wang, Junyao Yang, Zhichao Liu, Xiyang Wu, Anhao Li, Yue Yu, Ninghao Liu, Lichao Sun, Haotao Mi, Leowei Liang

TL;DR: A dense-reward terminal benchmark for measuring how far agents can progress on long, complex terminal workflows.

Paper Code

arXiv Preprint RST: Recursive Synthesis for Long-Horizon Terminal Tasks

Zhongzhi Li, Yucheng Shi, Zongxia Li, Ruhan Wang, Anhao Li, Zixun Huang, Junyao Yang, Lei Ke, Ninghao Liu, Haitao Mi, Leowei Liang

TL;DR: A recursive verified synthesis framework that scales long-horizon terminal-agent tasks, yielding 37,484 tasks at ~$0.05 each and lifting Qwen3.5 by up to 10 points across terminal benchmarks.

Paper

arXiv Preprint Harness Handbook: Making Evolving Agent Harnesses Readable, Navigable, and Editable

Ruhan Wang, Yucheng Shi, Zongxia Li, Zhongzhi Li, Yue Yu, Junyao Yang, Kishan Panaganti, Haitao Mi, Dongruo Zhou, Leoweiliang

TL;DR: A behavior-centric map and BGPD framework that help agents find code and plan harness edits.

Paper Project

EMNLP 2025 RewardDS: Privacy-Preserving Fine-Tuning for Large Language Models via Reward Driven Data Synthesis

Jianwei Wang, Chengming Shi, Junyao Yang, Haoran Li, Qianli Ma, Huiping Zhuang, Cen Chen, Ziqian Zeng^†

TL;DR: Using client-side reward models to filter synthetic data, mitigating noise while protecting privacy.

Paper Code

arXiv Preprint Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine

Junyao Yang*, Zongxia Li*, Yucheng Shi*, Zhongzhi Li, Ruhan Wang, Ruiwen Zhou, Haitao Mi, Shuicheng Yan^†, Leowei Liang

TL;DR: Alternating verified trajectory rewriting with critic-inheriting policy refinement, using turn-boundary credit assignment and stage-specific rewards to advance terminal agents from partial progress to strict task completion; Terminal-RSI-27B achieves 86.52% on Terminal-Bench 2.1 and 79.00% on Terminal-Bench Hard.

Paper Project Page

Tech Report (Core Contributor)

T1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks

Junyao Yang*^†, Yucheng Shi*, Zhongzhi Li, Ruhan Wang, Zongxia Li, Haitao Mi, Leowei Liang

TL;DR: A 122B MoE terminal agent trained by reinforcement learning — with stable MoE RL training framework and dense verification reward design aiming for Long-Horizon terminal tasks.

Paper Project Page Model

ACL 2026 ReasonAny: Incorporating Reasoning Capability to Any Model via Simple and Effective Model Merging

Junyao Yang, Chen Qian, Dongrui Liu^†, Wen Shen, Yong Liu^†, Jing Shao^†

TL;DR: Merging robust chain-of-thought capabilities into domain-specific models (Safety, Biomedicine) using Contrastive Gradient Identification.

Paper Code

AAAI 2026 RCP-Merging: Merging Long Chain-of-Thought Models with Domain-Specific Models by Considering Reasoning Capability as Prior

Junyao Yang, Jianwei Wang, Huiping Zhuang, Cen Chen, Ziqian Zeng*^†

TL;DR: Enhancing domain performance while preserving chain-of-thought reasoning abilities by treating reasoning as a prior.

Paper Code

arXiv Preprint Stale but Stable: Staleness-Adaptive Trust Regions for Stabilizing Asynchronous Reinforcement Learning

Junyao Yang, Yucheng Shi, Zongxia Li, Zhongzhi Li, Ruhan Wang, Xiangxin Zhou, Kishan Panaganti, Haitao Mi, Leowei Liang

TL;DR: Stabilizing asynchronous RL by adapting trust regions to rollout staleness, tightening high-mismatch updates while preserving ordinary-token behavior.

Paper

ACL 2025 PrivacyRestore: Privacy-Preserving Inference in Large Language Models via Privacy Removal and Restoration

Ziqian Zeng*^†, Jianwei Wang*, Junyao Yang*, Zhengdong Lu, Haoran Li, Huiping Zhuang, Cen Chen

TL;DR: Protecting privacy via activation steering using a protected meta-vector without retraining.

Paper Code

arXiv Preprint Long-Horizon-Terminal-Bench: Testing the Limits of Agents on Long-Horizon Terminal Tasks with Dense Reward-Based Grading

Zongxia Li, Zhongzhi Li, Yucheng Shi, Ruhan Wang, Junyao Yang, Zhichao Liu, Xiyang Wu, Anhao Li, Yue Yu, Ninghao Liu, Lichao Sun, Haotao Mi, Leowei Liang

TL;DR: A dense-reward terminal benchmark for measuring how far agents can progress on long, complex terminal workflows.

Paper Code

arXiv Preprint RST: Recursive Synthesis for Long-Horizon Terminal Tasks

Zhongzhi Li, Yucheng Shi, Zongxia Li, Ruhan Wang, Anhao Li, Zixun Huang, Junyao Yang, Lei Ke, Ninghao Liu, Haitao Mi, Leowei Liang

Paper

arXiv Preprint Harness Handbook: Making Evolving Agent Harnesses Readable, Navigable, and Editable

Ruhan Wang, Yucheng Shi, Zongxia Li, Zhongzhi Li, Yue Yu, Junyao Yang, Kishan Panaganti, Haitao Mi, Dongruo Zhou, Leoweiliang

TL;DR: A behavior-centric map and BGPD framework that help agents find code and plan harness edits.

Paper Project

Tech Report

AgentDoG: A Diagnostic Guardrail Framework for AI Agent Safety and Security

Shanghai Artificial Intelligence Laboratory (Contributor)

TL;DR: A state-of-the-art diagnostic guardrail framework utilizing a unified three-dimensional taxonomy to provide fine-grained monitoring and root-cause analysis of AI agent safety risks.

Paper Code Model #1 Paper of the day 机器之心

arXiv Preprint RSR: Scaling Trajectories for Complex Tasks through Recursive Self-Rewrite

Zongxia Li^†, Yucheng Shi^†, Zhongzhi Li^†, Junyao Yang, Ruhan Wang, Chengsong Huang, Fuxiao Liu, Haitao Mi, Jordan Boyd-Graber, Leowei Liang

Paper Model

arXiv Preprint SWR: Self-Supervised Scaling of Terminal Environments for Scientific Domains

Zhongzhi Li*, Yucheng Shi*, Zongxia Li*, Junyao Yang, Ruhan Wang, Yu Wang, Jingyuan Huang, Jichao Yu, Ninghao Liu^†, Haitao Mi, Leowei Liang

Paper

arXiv Preprint Entropy-Gradient Inversion: Moving Toward Internal Mechanism of Large Reasoning Models

Junyao Yang, Chen Qian, Kun Wang, Linfeng Zhang, Quanshi Zhang, Yong Liu, Dongrui Liu^†

TL;DR: Identifying Entropy-Gradient Inversion—a robust negative correlation between token entropy and logit gradients—as a geometric fingerprint of LRM reasoning capability, and proposing CorR-PO that embeds this signature into RL reward regularization to stabilize reasoning optimization without costly external verifiers.

Paper

arXiv Preprint The Why Behind the Action: Unveiling Internal Drivers via Agentic Attribution

Chen Qian, Peng Wang, Dongrui Liu^†, Junyao Yang, Dadi Guo, Ling Tang, Jilin Mei, Qihan Ren, Shuai Shao, Yong Liu, Jie Fu, Jing Shao, Xia Hu

TL;DR: A hierarchical framework for agentic attribution, using temporal likelihood and perturbation-based analysis to unveil internal factors driving LLM-based agent actions.

Paper Code

EMNLP 2025 RewardDS: Privacy-Preserving Fine-Tuning for Large Language Models via Reward Driven Data Synthesis

Jianwei Wang, Chengming Shi, Junyao Yang, Haoran Li, Qianli Ma, Huiping Zhuang, Cen Chen, Ziqian Zeng^†

TL;DR: Using client-side reward models to filter synthetic data, mitigating noise while protecting privacy.

Paper Code

Blogs

T1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks

TL;DR: A 122B MoE terminal agent trained by reinforcement learning — with stable MoE RL training framework and dense verification reward design aiming for Long-Horizon terminal tasks.

Stale but Stable: Staleness-Adaptive Trust Regions for Stabilizing Asynchronous Reinforcement Learning

TL;DR: Staleness analysis in asynchronous reinforcement learning: principled origins, diagnostics, and mitigations.

The Entropy-Gradient Inversion: A New Perspective on LLM Reasoning Capabilities

TL;DR: We discover that reasoning models exhibit a unique "fingerprint": a significant negative correlation between gradient strength and token entropy, which contradicts traditional base models. This capability emerges rapidly within the first 200 steps of SFT.