Research map / Multi-agent systems and games / Human-AI and goal-conditioned
World of Bits: An Open-Domain Platform for Web-Based Agents
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add supervised behavior cloning to random policy | MiniWoB agent · MiniWoB all envs | success rate: 20.8 -> 24.8 | Helped |
| fine-tune the SL policy with RL (A3C) | MiniWoB agent · MiniWoB all envs | success rate: 24.8 -> 34.8 | Helped |
| add RL after SL | MiniWoB agent · MiniWoB Click envs | success rate: 35.2 -> 48.5 | Helped |
| add RL after SL | MiniWoB agent · MiniWoB Drag envs | success rate: 24.7 -> 34.6 | Helped |
| add RL after SL | MiniWoB agent · MiniWoB Mouse+Keyboard envs | success rate: 3.5 -> 7.8 | Helped |
| add supervised learning on Mouse+Keyboard tasks vs random | MiniWoB agent · MiniWoB Mouse+Keyboard envs | success rate: 6.4 -> 3.5 | Hurt |
| add SL+RL on Compound tasks | MiniWoB agent · MiniWoB Compound envs | success rate: 3.3 -> 4.3 | No clear effect |
| add SL+RL | MiniWoB agent · MiniWoB % solved | % solved: 12 -> 26 | Helped |
| feed previously taken actions as input (ablation) | MiniWoB agent · MiniWoB | success rate | Hurt |
| weigh click and keyboard losses 10 times higher (ablation) | SL objective · MiniWoB | success rate | Helped |
| replace LocalCNN with GlobalCNN (ablation) | FormWoB/QAWoB agent · QAWoB | training loss: ∼0.2 -> ∼0.08 | Helped |
| replace LocalCNN with GlobalCNN (ablation) | web agent · QAWoB and FormWoB reward | reward | Helped |
Closest papers by meaning
- WebGym: Scaling Training Environments for Long-Horizon Visual Web Agents with Realistic Tasks (CVPR 2026 · similarity 0.89)
- WebFactory: Automated Compression of Foundational Language Intelligence into Grounded Web Agents (ICLR 2026 · similarity 0.89)
- WebArena: A Realistic Web Environment for Building Autonomous Agents (ICLR 2024 · similarity 0.88)
- WebRL: Training LLM Web Agents via Self-Evolving Online Curriculum Reinforcement Learning (ICLR 2025 · similarity 0.88)
- Web Agents with World Models: Learning and Leveraging Environment Dynamics in Web Navigation (ICLR 2025 · similarity 0.87)
- Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Use (ICML 2026 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.