Research map / Reinforcement learning / Visual and representation RL
You Only Live Once: Single-Life Reinforcement Learning
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 |
|---|---|---|---|
| Weight discriminator by Q-value (QWALE) instead of equal-weight GAIL-s | GAIL-s · Tabletop | avg steps to completion: 83.2k -> 44.4k | Helped |
| Weight discriminator by Q-value (QWALE) instead of equal-weight GAIL-s | GAIL-s · Pointmass | avg steps to completion: 100.9k -> 61.0k | Helped |
| Weight discriminator by Q-value (QWALE) instead of equal-weight GAIL-s | GAIL-s · Cheetah | avg steps to completion: 99.2k -> 74.3k | Helped |
| Weight discriminator by Q-value (QWALE) instead of equal-weight GAIL-s | GAIL-s · Kitchen | avg steps to completion: 111.3k -> 79.9k | Helped |
| Add distribution-matching reward (QWALE) instead of RND exploration bonus | SAC-RND · Pointmass | avg steps to completion: 184.1k -> 61.0k | Helped |
| Add RND exploration bonus to SAC fine-tuning | SAC · Tabletop | avg steps to completion: 123.7k -> 94.8k | Helped |
| Add RND exploration bonus to SAC fine-tuning | SAC · Cheetah | avg steps to completion: 128.9k -> 158.4k | Hurt |
| Pass action to discriminator (GAIL-sa vs GAIL-s) under dynamics shift | GAIL · Pointmass | avg steps to completion: 100.9k -> 140.4k | Hurt |
| Add behavioral cloning loss to SAC fine-tuning (ablation) | SAC · Pointmass | avg steps to completion: 198.1k -> 200.0k | Hurt |
| Add behavioral cloning loss to SAC fine-tuning (ablation) | SAC · Kitchen | avg steps to completion: 112.5k -> 101.1k | Helped |
| Train SAC from scratch instead of pretraining in source (ablation) | SAC · Tabletop | avg steps to completion: 123.7k -> 99.5k | Helped |
| Train SAC from scratch instead of pretraining in source (ablation) | SAC · Cheetah | avg steps to completion: 128.9k -> 200.0k | Hurt |
Closest papers by meaning
- A State-Distribution Matching Approach to Non-Episodic Reinforcement Learning (ICML 2022 · similarity 0.89)
- Autonomous Reinforcement Learning: Formalism and Benchmarking (ICLR 2022 · similarity 0.88)
- Continual Reinforcement Learning by Planning with Online World Models (ICML 2025 · similarity 0.85)
- Autonomous Option Invention for Continual Hierarchical Reinforcement Learning and Planning (AAAI 2025 · similarity 0.85)
- Discovering Multiple Solutions from a Single Task in Offline Reinforcement Learning (ICML 2024 · similarity 0.85)
- Learning to Reuse Policies in State Evolvable Environments (ICML 2025 · similarity 0.85)
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.