Research map / Reinforcement learning / Offline RL
DEAS: DEtached value learning with Action Sequence for Scalable Offline RL
Changyeon Kim, Haeone Lee, Younggyo Seo, Kimin Lee, Yuke Zhu
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 |
|---|---|---|---|
| use action sequences of length 8 instead of single-step actions (ablation) | DEAS · OGBench puzzle-4x4 | success rate (%): 21 -> 88 | Improved |
| increase action sequence length from 8 to 16 without enlarging the actor (ablation) | DEAS · OGBench puzzle-4x4 | success rate (%): 88 -> 51 | Hurt |
| enlarge actor to 1024x4 for H=16 (ablation) | DEAS · OGBench puzzle-4x4 | success rate (%): 51 -> 84 | Improved |
| add distributional HL-Gaussian classification loss on top of IQL detached value learning (ablation) | DEAS · OGBench puzzle-4x4 | success rate (%): 63 -> 88 | Improved |
| add IQL detached value learning on top of distributional HLG (ablation) | DEAS · OGBench puzzle-4x4 | success rate (%): 75 -> 88 | Improved |
| lower intra-option discount gamma1 from 0.99 to 0.9 (ablation) | DEAS · OGBench puzzle-4x4 | success rate (%): 81 -> 88 | Improved |
| lower gamma1 further from 0.9 to 0.8 (ablation) | DEAS · OGBench puzzle-4x4 | success rate (%): 88 -> 87 | No clear effect |
| enlarge critic network from 256x4 to 512x4 (ablation) | DEAS · OGBench puzzle-4x4 | success rate (%): 69 -> 88 | Improved |
| enlarge value network from 256x4 to 512x4 (ablation) | DEAS · OGBench puzzle-4x4 | success rate (%): 88 -> 50 | Hurt |
| increase n-step horizon in FQL without action sequence critics | FQL · OGBench scene-play (1M data) | success rate (%): 50 -> 36 | Hurt |
| use detached value learning with action sequences instead of coupled actor-critic (QC-FQL) | Action-sequence offline RL · OGBench puzzle-3x3 (1M data) | success rate (%): 62 -> 91 | Improved |
| use detached value learning with action sequences instead of coupled actor-critic (QC-FQL) | Action-sequence offline RL · OGBench cube-triple (10M data) | success rate (%): 83 -> 82 | No clear effect |
Closest papers by meaning
- Is Value Learning Really the Main Bottleneck in Offline RL? (NeurIPS 2024 · similarity 0.88)
- Video-Enhanced Offline Reinforcement Learning: A Model-Based Approach (ICML 2025 · similarity 0.87)
- Option-aware Temporally Abstracted Value for Offline Goal-Conditioned Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- Making Offline RL Online: Collaborative World Models for Offline Visual Reinforcement Learning (NeurIPS 2024 · similarity 0.87)
- Are Expressive Models Truly Necessary for Offline RL? (AAAI 2025 · similarity 0.87)
- Offline-Boosted Actor-Critic: Adaptively Blending Optimal Historical Behaviors in Deep Off-Policy RL (ICML 2024 · similarity 0.86)
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.
