Sitemap
A list of all the posts and pages found on the site. For you robots out there, there is an XML version available for digesting as well.
Pages
Posts
Habitat Engineering：AI时代，把判断力写进环境
Published:
当答案不再稀缺：AI时代，教育应该教什么？
Published:
🌐 An English version follows the Chinese version below.
Comprehensive LaTeX Template for Waseda University PhD Theses
Published:
Github: https://github.com/wanng-ide/phd_thesis_template_waseda_university
📚论文阅读
Published:
阅读的论文合集。
🤔一些思考
Published:
记录一些简单的思考💬。
Markdown Guide
Published:
📒 This page is from Academic Pages.
Academic Pages 简明双语指南
Published:
Quick Guide for Academic Pages (Chinese-English Bilingual)
从 视觉问答 Visual Question Answering（VQA）到 多模态表征 Multimodal representation learning 简单综述
Published:
文本主要是对VQA整个任务做一个综述。
portfolio
YUE开源音乐大模型
Open‑Sora
最大的开源的交错图文对数据集：PIN
PIN（Paired and INterleaved multimodal documents）
封神榜开源体系
一些自动脚本
记录一些简单的脚本项目
MNBVC超大规模中文语料集
A.S.E（AICGSecEval）开源仓库级代码评测体系
RoCo Challenge @ AAAI 2026 冠军：ARC-VLA（Anchor & Recovery Curriculum）
publications
⭐ [NTCIR 15] SKYMN at the NTCIR-15 DialEval-1 Task
Published in NTCIR, 2020-12
Dialogue Quality, BiLSTM + Attention), CNN (Convolutional Neural Network), Pre‑trained Language Model, MoE
⭐ [EMNLP 2021] MIRTT: Learning Multimodal Interaction Representations from Trilinear Transformers for Visual Question Answering
Published in EMNLP, 2021-10
Multimodal Interaction, Trilinear Transformer, Visual Question Answering (VQA), Two-Stage Workflow
⭐ Towards No.1 in CLUE Semantic Matching Challenge: Pre-trained Language Model Erlangshen with Propensity-Corrected Loss
Published in Arxiv, 2022-08
Semantic Matching, Pre-trained Language Model (PLM), Propensity‑Corrected Loss (PCL), LUE Semantic Matching Challenge
⭐ Fengshenbang 1.0: Being the Foundation of Chinese Cognitive Intelligence
Published in Arxiv, 2022-09
Chinese Pre-trained Models, Foundation Models, Open-Source Ecosystem, Cognitive Intelligence
[ACM 2022] Breaking Isolation: Multimodal Graph Fusion for Multimedia Recommendation by Edge-wise Modulation
Published in ACM MM, 2022-10
Multimedia Recommendation, Graph Fusion, Edge-wise Modulation, Graph Convolutional Network (GCN)
⭐ [EMNLP 2022] Zero-Shot Learners for Natural Language Understanding via a Unified Multiple Choice Perspective
Published in EMNLP, 2022-10
Zero‑Shot Learning, Multiple Choice Format, Pre‑trained Masked Language Model (PMLM), Unified Multiple Choice model
⭐ [CVPR 2023] MAP: Modality-Agnostic Uncertainty-Aware Vision-Language Pre-training Model
Published in CVPR, 2023-06
Multimodality, Uncertainty Modeling, Vision-Language Pre-training, Probability Distribution Encoder (PDE)
⭐ [ACL 2023] Solving Math Word Problems via Cooperative Reasoning induced Language Models
Published in ACL, 2023-07
System 1 & System 2, Cooperative Reasoning (CoRe),Stepwise Feedback, Math Word Problems
[ACL 2023] UniEX: An Effective and Efficient Framework for Unified Information Extraction via a Span-extractive Perspective
Published in ACL, 2023-07
Information Extraction , Unified Across IE Tasks, Triaffine Attention, Span‑extractive Framework, Low‑resource Transferability
GameGPT: Multi-agent Collaborative Framework for Game Development
Published in Arxiv, 2023-10
Game Development, Multi-agent Collaboration, LLM, Redundancy
⭐ [SIGIR-AP 2023] EALM: Introducing Multidimensional Ethical Alignment in Conversational Information Retrieval
Published in SIGIR-AP, 2023-11
Multidimensional Ethics, Ethical Judgment, Large Multimodal Models (LMMs), Binary & Multi‑label Classification
CMMMU: A Chinese Massive Multi-discipline Multimodal Understanding Benchmark
Published in Arxiv, 2024-01
Multimodal Understanding, Subject Knowledge Reasoning , University Exam Questions, LMM Performance Evaluation
[COLM 2024] StructLM: Towards Building Generalist Models for Structured Knowledge Grounding
Published in COLM, 2024-02
Structured Knowledge Grounding (SKG), Instruction tuning, Generalist model, Tables / Graphs / Databases
⭐ [ACM TOIS 2024] SSR: Solving Named Entity Recognition Problems via a Single-stream Reasoner
Published in ACM TOIS, 2024-05
Named Entity Recognition (NER), Machine Reading Comprehension (MRC), Single-Stream Reasoner (SSR), Multi-choice Input Format
⭐🚩 PIN: A Knowledge-Intensive Dataset for Paired and Interleaved Multimodal Documents
Published in Arxiv, 2024-06
Knowledge‑intensive, Paired and Interleaved Documents, Multimodal Datasets, Data Format
⭐ [SIGIR-AP 2024 最佳论文提名] Data-Efficient Massive Tool Retrieval: A Reinforcement Learning Approach for Query-Tool Alignment with Language Models
Published in SIGIR-AP, 2024-10
Massive Tool Retrieval (MTR), Query‑Tool Alignment (QTA), Massive Tool Retrieval Benchmark, DPO
⭐ [EMNLP 2024] ToolBeHonest: A Multi-level Hallucination Diagnostic Benchmark for Tool-Augmented Large Language Models
Published in EMNLP, 2024-11
ToolBeHonest, Hallucination, LLM, Multi‑level Diagnostic
[EMNLP 2024] HoLLMwood: Unleashing the Creativity of Large Language Models in Screenwriting via Role Playing
Published in EMNLP, 2024-11
Screenwriting, Large Language Models (LLMs), Role Playing, Creative Generation, Multi-Agent Collaboration
⭐ [ICLR 2025] ChartMimic: Evaluating LMM’s Cross-Modal Reasoning Capability via Chart-to-Code Generation
Published in ICLR, 2025-04
Large Multimodal Models (LMMs), Chart Understanding, Code Generation, Cross-modal Reasoning
🚩 [ACL 2025] Chain-of-Reasoning: Towards Unified Mathematical Reasoning in Large Language Models via a Multi-Paradigm Perspective
Published in ACL, 2025-07
Multi-Paradigm, Large Language Model (LLM), Progressive Paradigm Training (PPT), Zero-shot Generalization
💭 VeriGUI: Verifiable Long-Chain GUI Dataset
Published in Arxiv, 2025-08
Long-Chain Complexity, GUI Agents, Subtask-Level Verifiability, POMDP
💭 Nav-R^2 Dual-Relation Reasoning for Generalizable Open-Vocabulary Object-Goal Navigation
Published in Arxiv, 2025-12
Object-Goal Navigation, Chain-of-Thought, Dual-Relation Reasoning, Similarity-Aware Memory
🚩 [AAAI 2026 (Oral)] O-DisCo-Edit: Object Distortion Control for Unified Realistic Video Editing
Published in AAAI, 2026-01
Video Editing, Diffusion Models, Unified Control Signal, Object Distortion Control
💭🚩 OneLatent: Single-Token Compression for Visual Latent Reasoning
Published in Arxiv, 2026-02
Latent Reasoning, Chain-of-Thought Compression, Visual Supervision, DeepSeek-OCR, Minimum Description Length, Reasoning Efficiency
🚩 [TMLR 2026] A Survey of Reasoning in Autonomous Driving Systems: Open Challenges and Emerging Paradigms
Published in TMLR, 2026-03
Autonomous Driving, Reasoning, LLM/MLLM, Cognitive Hierarchy, Long-tail Scenarios, Social Game, Survey
[ICLR 2026] YuE: Scaling Open Foundation Models for Long-Form Music Generation
Published in ICLR, 2026-04
Long-form music (full-song) generation, Audio Language Model, Music In-Context Learning, Controllability, human preference evaluation
💭🚩 Think-with-Rubrics: From External Evaluator to Internal Reasoning Guidance
Published in Arxiv, 2026-05
Instruction Following, Rubric-Guided Reasoning, Rubric-as-Reward, Self-Generated Rubrics, Reinforcement Learning, Self-Consistency
💭 Velocity-Space 3D Asset Editing
Published in Arxiv, 2026-05
3D Asset Editing, Rectified Flow, Velocity-Space Editing, Training-Free Editing, Mask-Free Editing, TRELLIS
[ICML 2026] Smaller Models are Natural Explorers for Policy-Level Diversity in GRPO
Published in ICML, 2026-06
GRPO, Policy-Level Diversity, Small-to-Large Policy Optimization, RLVR, Rollout Diversity, Progressive Annealing
💭⭐ WebRISE: Requirement-Induced State Evaluation for MLLM-Generated Web Artifacts
Published in Arxiv, 2026-06
MLLM-Generated Web Artifacts, Requirement-Induced State Evaluation, Interaction Contract Graph, Web Generation Benchmark, DOM/Visual Assertion
[ACL 2026] A.S.E: A Repository-Level Benchmark for Evaluating Security in AI-Generated Code
Published in ACL, 2026-07
AI-Generated Code Security, Repository-Level Benchmark, Large Language Models (LLMs), Common Vulnerabilities and Exposures (CVEs)
🚩 [ACL 2026] ProFit: Leveraging High-Value Signals in SFT via Probability-Guided Token Selection
Published in ACL, 2026-07
Probability-Guided Token Selection, High-Value Semantic Signals, Gradient Interference, Training Stability & Rapid Convergence
💭 Route by Kinematics, Act by Observation: Kinematics-Supervised Expert Routing in MoE-Augmented VLA
Published in Arxiv, 2026-07
Vision-Language-Action, Mixture-of-Experts, Kinematics-Supervised Routing, Robot Manipulation, DIYRobot
💭🚩 Simple-OPD: Demystifying Warm-up for On-policy Distillation
Published in Arxiv, 2026-08
On-Policy Distillation, Warm-up, Chain-of-Thought, Teacher Compatibility, LoRA, Reasoning Models
🚩 AutoResearch: Insight In, Hallucination Out
Published in Arxiv, 2026-08
Autonomous Research, Multi-Agent Systems, Idea Generation, Evidence-Grounded Execution, Independent Verification, Scientific Discovery
⭐🚩 Learning Where Outcomes Change: Credit-Addressable Reasoning for Multimodal Geometry
Published in Arxiv, 2026-08
Multimodal Geometry Reasoning, Credit Assignment, Code-CoT, CE-GRPO, Executable Perception Code, Event-Level Reinforcement Learning
[EMNLP 2026] AnyCap Project: A Unified Framework, Dataset, and Benchmark for Controllable Omni-modal Captioning
Published in EMNLP, 2026-10
Controllable Captioning, Omni-modal Intelligence, Plug-and-play Framework, Multimodal Large Language Models
⭐ [EMNLP 2026] SIN-Bench: Tracing Native Evidence Chains in Long-Context Multimodal Scientific Interleaved Literature
Published in EMNLP, 2026-10
Multimodal Large Language Models, Long-Context Understanding, Evidence Chain, Scientific Literature, Benchmark
🚩 [EMNLP 2026] Internalize the Temperature: On-Policy Self-Distillation as Policy Reheater for Reinforcement Learning
Published in EMNLP, 2026-10
RLVR, Entropy Collapse, Policy Reheating, On-Policy Self-Distillation, Temperature Scaling, Continued Reinforcement Learning
talks
IDEA研究院原作团队解读封神榜体系：致力于成为中文认知智能的基础设施
Published:
随着大模型在自然语言处理、计算机视觉等多个领域兴起，认知智能正在经历范式上的变化。借助大规模的数据以及庞大的参数量，这些模型展现出能够有效处理各种任务的特征，并正在以惊人的速度被部署到各个专业领域中，对社会和经济发展产生深远的影响。但是目前中文社区出现了某种停滞不前的现象，因为模型的体量已经从原本的百万参数飞跃至千亿级别，一些高校和传统公司并不具备足够的算力，也缺少有效的基础设施帮助他们训练和使用模型。因此，要推动人工智能技术进一步发展，坚实的基础设施尤为重要。
首个中文Stable Diffusion模型背后的技术：IDEA研究院封神榜团队揭秘
Published:
全面讲解太乙系列模型从模型的生产到应用。该分享从训练，微调和加速等角度揭秘封神榜开源体系之一的太乙系列（多模态系列）模型是如何生产的。基于该团队训练后开源的权重，讲解如何推理加速以及如何部署在 webui 和 dreambooth 等应用上。
Large Models bridge the Digital-Real World Gap: from Understanding to Generation
Published:
NCAA 2023 tutorial speaker
SIGIR-AP 2023 in Beijing, China
Published:
EALM: Introducing Multidimensional Ethical Alignment in Conversational Information Retrieval
ACL 2025 in Vienna, Austria
Published:
A presentation and a poster.
SIGIR-AP 2025 in Xi’an, China
Published:
From December 7–10, 2025, SIGIR-AP 2025 was held in Xi’an, China. As one of the organizers, I co-organized the BREV-RAG workshop — Beyond Relevance-based EValuation of RAG systems — at SIGIR-AP 2025.
teaching
Teaching experience 1
Undergraduate course, University 1, Department, 2014
This is a description of a teaching experience. You can use markdown like any other post.
Teaching experience 2
Workshop, University 1, Department, 2015
This is a description of a teaching experience. You can use markdown like any other post.