Awesome Lists
Curated open-source research maps maintained under OpenEnvision, spanning multimodal modeling and visual-agent systems.
OpenEnvision
A structured survey for multimodal models, covering MLLMs, unified multimodal models, native multimodal models, and adjacent awesome lists.
Scope
Image-text centered multimodal modeling with annotations for video, audio, 3D, and omni extensions.
Focus
Architecture-first categorization, model taxonomy, and links to papers, code, demos, and related lists.
A curated research index for visual agents that perceive, ground, plan, act, create, and evaluate in visually grounded environments.
Scope
GUI agents, computer-use agents, embodied VLA systems, visual reasoning loops, and visual-agent safety.
Focus
Research taxonomy, reading pathways, benchmarks, environments, tools, and engineering resources.
Codebase
Open-source codebases I lead or contribute to, supporting reproducible research infrastructure and community use.
OpenEnvision
An open-source WorldFoundry project from OpenEnvision, supporting research and development toward open world intelligence.
Purpose
Builds an open foundation for developing, exploring, and sharing world-model research.
Collaboration
Developed under OpenEnvision with an emphasis on open-source collaboration and reproducible research.
OpenDCAI
An open-source framework for generating games and assets with coding agents, covering images, 3D assets, motion, audio, and cinematic video across multiple game engines.
Capabilities
Combines asset generation with gameplay and UI code generation to build playable games from requirements.
Supported Engines
Supports game construction with Unreal Engine 5, Unity, Godot 4, Blender, and three.js.
Participated in OpenDCAI's unified world-model codebase, which standardizes model invocation, documents advanced world-model research, and integrates reusable pipelines for perception, reasoning, generation, and interaction.
Role
Participated as a contributor to the project rather than as the primary lead.
Focus
Unified framework design, method integration, documentation, and reusable scripts for advanced world-model research.
Optimizer
Optimizer research and training systems where I participated as a contributor.
xie-lab-ml
Participated in Mano and ManoV2, optimizer research for restriking manifold optimization in LLM training.
Role
Participated as a contributor to the optimizer project.
Focus
Restriking manifold optimization methods and training-oriented optimizer development for large language models.
Knowledge Sharing Community
Community infrastructure for discovering, sharing, and curating high-signal AI research writing and optimization resources.
OpenEnvision
Knowledge-sharing community and discovery platform for curated AI research blogs, helping researchers find durable technical writing across models, agents, safety, efficiency, and world models.
Purpose
Curates high-quality AI research blogs and technical insights for researchers and builders.
Signals
Highlights GitHub stars as a simple signal of community interest.
Knowledge-sharing community for large-scale optimization, helping researchers discover optimizer algorithms, compare training strategies, and follow practical resources for efficient deep learning.
Purpose
Collects optimizer-centric resources for large-scale machine learning, including algorithms, papers, tutorials, and implementation references.
Signals
Highlights GitHub stars as a simple signal of repository interest.