Research map / Reinforcement learning / Offline RL
Improving Zero-Shot Generalization in Offline Reinforcement Learning using Generalized Similarity Functions
Bogdan Mazoure, Ilya Kostrikov, Ofir Nachum, Jonathan Tompson
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 reward-based GVF quantile similarity with GSF on CQL | CQL · Offline Procgen, climber | standardized score: -0.263158 -> 1.894737 | Improved |
| Use reward-based GVF quantile similarity with GSF on CQL | CQL · Offline Procgen, leaper | standardized score: 0.497143 -> -0.091429 | Hurt |
| Use action-count cumulants instead of reward or successor-feature cumulants (ablation) | GSF · Offline Procgen | zero-shot performance | Mixed |
| Use categorical cross-entropy multiclass labels instead of pairwise InfoNCE (ablation) | GSF · Offline Procgen | contrastive loss error | Improved |
| Vary quantile-bin count K (ablation) | GSF · Offline Procgen | zero-shot performance | Mixed |
| Vary softmax temperature tau (ablation) | GSF · Offline Procgen | zero-shot performance | Mixed |
| Vary the structure of projection network h (ablation) | GSF · Offline Procgen | zero-shot performance | Mixed |
Closest papers by meaning
- Provable Zero-Shot Generalization in Offline Reinforcement Learning (ICML 2025 · similarity 0.92)
- Doubly Mild Generalization for Offline Reinforcement Learning (NeurIPS 2024 · similarity 0.91)
- The Generalization Gap in Offline Reinforcement Learning (ICLR 2024 · similarity 0.91)
- Zero Shot Generalization of Vision-Based RL Without Data Augmentation (ICML 2025 · similarity 0.90)
- Zero-Shot Reinforcement Learning from Low Quality Data (NeurIPS 2024 · similarity 0.90)
- Scaling Offline Model-Based RL via Jointly-Optimized World-Action Model Pretraining (ICLR 2025 · similarity 0.89)
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.
