GraphVerse

Open Science · Graph World Models

Where graph structure
meets world dynamics.

GraphVerse is the world's first open academic exchange and collaboration platform dedicated to Graph World Models — uniting structured topological representation with dynamical simulation to power the next generation of AI.

  • Physical Simulation
  • Embodied AI
  • AI for Science
  • Complex Networks

Mission & Vision

Closing the gap between structure and dynamics

We are dedicated to bridging the technical gap between structured topological representation (Graph Neural Networks) and dynamical reasoning (World Models). By bringing together world-class academic resources, frontier literature and open-source tools, we drive theoretical breakthroughs and real-world deployment of Graph World Models.

GNN

Graph Neural Networks

Relational, compositional representations of entities and their interactions.

+ WM

World Models

Learned dynamics that predict, imagine and plan how systems evolve over time.

= GWM

Graph World Models

Structured, interpretable simulators that generalize across scales and domains.

Key Modules

One hub for the entire research workflow

01

Literature & Code Hub

Systematically categorizes and tracks the latest results from NeurIPS, ICLR, ICML and arXiv. Following Open Science principles, we host no paper files — only public links to arXiv, official journals and official GitHub repositories.

  • NeurIPS
  • ICLR
  • ICML
  • arXiv
  • GitHub

02

Academic Activities & Feeds

Aggregates top-conference workshops, online reading-group announcements and Call for Papers news, plus a daily feed of newly published papers powered by the arXiv API.

  • Workshops
  • Reading Groups
  • CFP
  • Daily arXiv

03

Interactive Playground & Tools

Structured state-evolution demos, domain benchmark datasets and reference implementations in mainstream graph-learning frameworks — a one-stop navigator for research and development.

  • PyTorch Geometric
  • DGL
  • JAX
  • Benchmarks

Commercial Potential

Frontier research with industrial-scale impact

Graph World Models show exceptional promise wherever systems are made of interacting parts that evolve over time.

Embodied Intelligence

Structured world models for robot perception, manipulation and control.

Industrial Simulation

High-fidelity surrogates for complex industrial and physical systems.

Drug & Molecular Design

Graph-native dynamics for molecules, proteins and biomedical discovery.

Transport & Energy Networks

Forecasting and optimal control of traffic, grid and infrastructure networks.

Our Ecosystem

A closed loop from research to capital

  1. 01

    Research

    Frontier theory & open literature

  2. 02

    Technology

    Tools, benchmarks & open source

  3. 03

    Industry

    Deployment in real-world systems

  4. 04

    Capital

    Investment that accelerates the loop

Investment & Collaboration

Build the future of Graph World Models with us

We warmly invite colleagues across academia and industry worldwide to engage in deep collaboration.

Academia & Community

Academic exchange & community building

Research teams are welcome to submit their latest work, co-host online and offline workshops, or contribute to the platform's open-source development.

  • Submit papers & code
  • Co-host workshops & reading groups
  • Contribute to open source
Get in touch
Industry & Capital

Industrial deployment & capital empowerment

We are actively seeking strategic partnerships and early-stage investment from venture capital firms, industrial funds and corporate R&D labs to jointly accelerate the commercialization of frontier technology.

  • Venture capital & early-stage investment
  • Industrial & strategic funds
  • Corporate R&D lab partnerships
Get in touch

Contact Us

Let's talk.

Interested in frontier Graph World Model research, or have ideas for technical exchange, paper recommendations, project collaboration or investment? We'd love to hear from you.

venslu.pro@gmail.com
Send an email