Built with and by Teycir Ben Soltane•
How to Use•FAQ•GitHub•arXiv.org•
Share:
ArXivCSExplorer
☆☆Bookmarks🏆RSSHow to UseFAQ
Home/Authors/Peng Wang

Peng Wang

12 indexed papers

Recent (6 mo)
12
With code
0
Influential cites
0
Benchmarked
0

Publications per year

12
26

Top categories

NLP×8AI×7Crypto×3Vision×3ML×3Comp. Eng.×2Architecture×1Distributed×1

Frequent co-authors

Yanqiao Zhu3×
Wupeng Wang3×
Xipeng Qiu3×
Kai Yu3×
Zhifu Gao3×
Xiangang Li3×

Research Timeline

2026
FinBoardBench: Benchmarking Dynamic Wealth Management and Strategic Financial Reasoning of LLMs via Board Game Simulations

The paper introduces FinBoardBench, a novel evaluation suite using financial board games to demonstrate that current LLMs, despite strong static reasoning, fail at complex, dynamic wealth management and strategic decision-making.

FinBoardBench: Benchmarking Dynamic Wealth Management and Strategic Financial Reasoning of LLMs via Board Game Simulations

The paper introduces FinBoardBench, a novel evaluation suite using financial board games to demonstrate that current LLMs, despite strong static reasoning, fail at complex, dynamic wealth management and strategic decision-making.

AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security

The paper introduces AgentDoG 1.5, a lightweight and scalable alignment framework that significantly improves AI agent safety and security for complex, open-world agentic scenarios.

Towards Human-Like Interactive Speech Recognition With Agentic Correction and Semantic Evaluation

The paper proposes Agentic ASR, a closed-loop framework that treats ASR as a multi-turn refinement task, significantly improving semantic accuracy over traditional token-level metrics.

AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security

The paper introduces AgentDoG 1.5, a lightweight and scalable alignment framework that significantly improves AI agent safety and security for complex, open-world agentic scenarios.

Towards Human-Like Interactive Speech Recognition With Agentic Correction and Semantic Evaluation

The paper proposes Agentic ASR, a closed-loop framework that treats ASR as a multi-turn refinement task, significantly improving semantic accuracy over traditional token-level metrics.

AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security

The paper introduces AgentDoG 1.5, a lightweight and scalable alignment framework that significantly improves AI agent safety and security for complex open-world agent deployments.

eMoT: evolving Memory-of-Thought via Symbolic Anchoring and Memory Corrosion

The eMoT framework enhances multi-step reasoning in LLMs by treating reasoning as an evolving memory, stabilizing performance through symbolic computation and structured refinement.

eMoT: evolving Memory-of-Thought via Symbolic Anchoring and Memory Corrosion

The eMoT framework enhances multi-step reasoning in LLMs by treating reasoning as an evolving memory, stabilizing performance through symbolic computation and structured refinement.

Verifiable Environments Are LEGO Bricks: Recursive Composition for Reasoning Generalization

This paper introduces RACES, a framework for recursively assembling verifiable environments to enhance Large Language Model reasoning capabilities.

GigaSpeechBench: A Real-World Multilingual Speech-to-Text Benchmark

The paper introduces GigaSpeechBench, a comprehensive multilingual and multidimensional ASR & AST benchmark with 680 hours of human-annotated speech, featuring 12 low-resource languages, 6 Chinese dialects, 6 English accents, dense terminology, older adult and child speech, and human-annotated translations.

ZipFusion: Fusing Lossless KV Decompression with Attention for Efficient LLM Decoding

This paper presents ZipFusion, a method that enables attention to consume losslessly compressed key-value (KV) cache directly, improving model decode throughput and reducing memory-bandwidth bound.

Highlighted terms show continued research focus across papers

Papers

cs.ARcs.DCEmpiricalRecentOct 24, 2026

ZipFusion: Fusing Lossless KV Decompression with Attention for Efficient LLM Decoding

Yuetao Chen, Peng Wang, Hong Xu

This paper presents ZipFusion, a method that enables attention to consume losslessly compressed key-value (KV) cache directly, improving model decode throughput and reducing memory-bandwidth bound.

View →
eess.ASEmpirical
Recent
Jun 27, 2026

GigaSpeechBench: A Real-World Multilingual Speech-to-Text Benchmark

Yujie Tu, Yifan Yang, Tianrui Wang, Yanqiao Zhu +32 more

The paper introduces GigaSpeechBench, a comprehensive multilingual and multidimensional ASR & AST benchmark with 680 hours of human-annotated speech, featuring 12 low-resource languages, 6 Chinese dia…

View →
cs.CLEmpiricalRecentJun 10, 2026

Verifiable Environments Are LEGO Bricks: Recursive Composition for Reasoning Generalization

Hao Xiang, Qiaoyu Tang, Le Yu, Yaojie Lu +7 more

This paper introduces RACES, a framework for recursively assembling verifiable environments to enhance Large Language Model reasoning capabilities.

View →
cs.AIRecentJun 1, 2026

eMoT: evolving Memory-of-Thought via Symbolic Anchoring and Memory Corrosion

Xiang Li, Jiwei Wei, Ke Liu, Yitong Qin +4 more

The eMoT framework enhances multi-step reasoning in LLMs by treating reasoning as an evolving memory, stabilizing performance through symbolic computation and structured refinement.

View →
cs.AIRecentJun 1, 2026

eMoT: evolving Memory-of-Thought via Symbolic Anchoring and Memory Corrosion

Xiang Li, Jiwei Wei, Ke Liu, Yitong Qin +4 more

The eMoT framework enhances multi-step reasoning in LLMs by treating reasoning as an evolving memory, stabilizing performance through symbolic computation and structured refinement.

View →
cs.AIcs.CLcs.CRRecentMay 28, 2026

AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security

Dongrui Liu, Yu Li, Zhonghao Yang, Peng Wang +46 more

The paper introduces AgentDoG 1.5, a lightweight and scalable alignment framework that significantly improves AI agent safety and security for complex, open-world agentic scenarios.

View →
cs.AIcs.CLRecentMay 28, 2026

Towards Human-Like Interactive Speech Recognition With Agentic Correction and Semantic Evaluation

Zixuan Jiang, Yanqiao Zhu, Peng Wang, Qinyuan Chen +7 more

The paper proposes Agentic ASR, a closed-loop framework that treats ASR as a multi-turn refinement task, significantly improving semantic accuracy over traditional token-level metrics.

View →
cs.AIcs.CLcs.CRRecentMay 28, 2026

AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security

Dongrui Liu, Yu Li, Zhonghao Yang, Peng Wang +46 more

The paper introduces AgentDoG 1.5, a lightweight and scalable alignment framework that significantly improves AI agent safety and security for complex, open-world agentic scenarios.

View →
cs.AIcs.CLRecentMay 28, 2026

Towards Human-Like Interactive Speech Recognition With Agentic Correction and Semantic Evaluation

Zixuan Jiang, Yanqiao Zhu, Peng Wang, Qinyuan Chen +7 more

The paper proposes Agentic ASR, a closed-loop framework that treats ASR as a multi-turn refinement task, significantly improving semantic accuracy over traditional token-level metrics.

View →
cs.AIcs.CLcs.CRRecentMay 28, 2026

AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security

Dongrui Liu, Yu Li, Zhonghao Yang, Peng Wang +46 more

The paper introduces AgentDoG 1.5, a lightweight and scalable alignment framework that significantly improves AI agent safety and security for complex open-world agent deployments.

View →
cs.CLcs.CERecentMay 27, 2026

FinBoardBench: Benchmarking Dynamic Wealth Management and Strategic Financial Reasoning of LLMs via Board Game Simulations

Xuesi Hu, Peng Wang, Jinpeng Miao, Xilin Tao +6 more

The paper introduces FinBoardBench, a novel evaluation suite using financial board games to demonstrate that current LLMs, despite strong static reasoning, fail at complex, dynamic wealth management a…

View →
cs.CLcs.CERecentMay 27, 2026

FinBoardBench: Benchmarking Dynamic Wealth Management and Strategic Financial Reasoning of LLMs via Board Game Simulations

Xuesi Hu, Peng Wang, Jinpeng Miao, Xilin Tao +6 more

The paper introduces FinBoardBench, a novel evaluation suite using financial board games to demonstrate that current LLMs, despite strong static reasoning, fail at complex, dynamic wealth management a…

View →