Research map / Multi-agent systems and games / Human-AI and goal-conditioned
What You Think is What You See: Driving Exploration in VLM Agents via Visual-Linguistic Curiosity
Haoxi Li, Qinglin Hou, Jianfei Ma, Jinxiang Lai, Tao Han, Sikai Bai, Jingcai Guo, Jie ZHANG, Song Guo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 GLANCE visual-linguistic curiosity | VAGEN-Full · Overall (5 tasks) | overall score: 0.81 -> 0.86 | Improved |
| add GLANCE curiosity | VAGEN-Full · Sokoban | success rate: 0.79 -> 0.85 | Improved |
| add GLANCE curiosity | VAGEN-Base · Sokoban | success rate: 0.61 -> 0.74 | Improved |
| add GLANCE curiosity | VAGEN-Base · Overall (5 tasks) | overall score: 0.76 -> 0.80 | Improved |
| add GLANCE curiosity | Turn-level PPO · PrimitiveSkill (average) | success rate: 0.25 -> 0.66 | Improved |
| add GLANCE curiosity | Turn-level PPO · Overall (5 tasks) | overall score: 0.55 -> 0.69 | Improved |
| add GLANCE curiosity | VAGEN-Full · PrimitiveSkill (average) | success rate: 0.97 -> 0.97 | No clear effect |
| add GLANCE curiosity | Turn-level PPO · Navigation (average) | success rate: 0.81 -> 0.79 | No clear effect |
| add dense reasoning reward r_reason (ablation) | VAGEN-Full · Sokoban | success rate: 0.63 -> 0.79 | Improved |
| add dense reasoning reward r_reason (ablation) | VAGEN-Full · FrozenLake | success rate: 0.70 -> 0.72 | Improved |
| add GLANCE curiosity without reasoning reward (ablation) | VAGEN-Full w/o r_reason · Sokoban | success rate: 0.63 -> 0.67 | Improved |
| add GLANCE curiosity without reasoning reward (ablation) | VAGEN-Full w/o r_reason · Navigation (avg) | success rate: 0.79 -> 0.81 | Improved |
Closest papers by meaning
- VAGEN: Reinforcing World Model Reasoning for Multi-Turn VLM Agents (NeurIPS 2025 · similarity 0.91)
- SenseSearch: Empowering Vision-Language Models with High-Resolution Agentic Search-Reasoning via Reinforcement Learning (CVPR 2026 · similarity 0.89)
- Expanding LLM Agent Boundaries with Strategy-Guided Exploration (NeurIPS 2026 · similarity 0.89)
- Code as Reward: Empowering Reinforcement Learning with VLMs (ICML 2024 · similarity 0.88)
- SENSEI: Semantic Exploration Guided by Foundation Models to Learn Versatile World Models (ICML 2025 · similarity 0.88)
- Scaling Agentic Reinforcement Learning for Tool-Integrated Reasoning in VLMs (CVPR 2026 · similarity 0.88)
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.
