Research map / Reinforcement learning / Visual and representation RL
Improving Intrinsic Exploration with Language Abstractions
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 language goals (L-AMIGo) in place of coordinate goals | AMIGo · 13 tasks overall | IQM reward: .27 absolute, 47% relative | Helped |
| add language novelty bonus (L-NovelD) | NovelD · 13 tasks overall | IQM reward: .35 absolute, 85% relative | Helped |
| naive fixed reward for any message instead of novelty/difficulty-based language reward (ablation) | IMPALA · MiniGrid and MiniHack | reward | Hurt |
| replace language goals/messages with one-hot ids (hide semantics) in L-AMIGo (ablation) | L-AMIGo · 13 tasks overall | IQM reward: .20 absolute, 32% relative drop vs language | Hurt |
| replace language messages with one-hot ids in L-NovelD (ablation) | L-NovelD · 13 tasks overall | IQM reward | No clear effect |
| add language to NovelD on larger environments | NovelD · KeyCorridorS5R3 | IQM reward: 0.00 -> 0.88 | Helped |
| add language goals to AMIGo | AMIGo · WoD-Medium | IQM reward: 1.00 -> 1.00 | No clear effect |
| add language goals to AMIGo on hardest task | AMIGo · WoD-Hard | IQM reward: 0.00 -> 0.85 | Helped |
| encode language into the state representation for MiniGrid agents (ablation) | MiniGrid baselines · MiniGrid | performance | Helped |
| add language to NovelD on hardest MultiRoom | NovelD · MultiRoom-N4-Extreme | IQM reward: 0.00 -> 0.00 | No clear effect |
Closest papers by meaning
- Representation-Based Exploration for Language Models: From Test-Time to Post-Training (ICLR 2026 · similarity 0.90)
- Expanding LLM Agent Boundaries with Strategy-Guided Exploration (NeurIPS 2026 · similarity 0.89)
- Prompted Policy Search: Reinforcement Learning through Linguistic and Numerical Reasoning in LLMs (NeurIPS 2025 · similarity 0.88)
- SENSEI: Semantic Exploration Guided by Foundation Models to Learn Versatile World Models (ICML 2025 · similarity 0.88)
- Ground-Compose-Reinforce: Grounding Language in Agentic Behaviours using Limited Data (NeurIPS 2025 · similarity 0.88)
- DISCOVER: Automated Curricula for Sparse-Reward Reinforcement Learning (NeurIPS 2025 · 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.