VisualTreeSearch Playground
Welcome to the playground for VisualTreeSearch, an intuitive interface for understanding web agent decision processes. Explore and interact with the demo presented in the paper.
Available Algorithms
Explore different tree search algorithms and their applications in web agent decision making
Monte Carlo Tree Search
A heuristic search algorithm that intelligently builds decision trees by balancing exploration of untested paths with exploitation of promising strategies
Language Agent Tree Search
A search algorithm that enhances language models by integrating tree search techniques, enabling systematic exploration of decision paths with environmental feedback to unify reasoning, acting, and planning capabilities.
BFS/DFS
Classic graph traversal algorithms: Breadth-First Search for level-by-level exploration and Depth-First Search for deep path exploration.
Project Architecture
A modern stack combining powerful frontend technologies with a robust Python backend
- FrameworkNext.js
- UI LibraryTailwind CSS
- State ManagementReact Hooks
- StylingCSS Modules
- FrameworkFastAPI
- DeploymentAWS ECS
- ContainerDocker
- RuntimePython 3.9+
- DNS ManagementAWS Route 53
- SSL CertificatesAWS ACM
- CDNCloudflare
- MonitoringAWS CloudWatch
This architecture combines modern frontend technologies with a Python-based backend, using cloud services for deployment and security.