Zero-Search Reinforcement Learning for Slay the Spire 2 with Failure-Driven Restarts
cs.LG, cs.AI
Submitted: 2026-11-09
Updated: 2026-11-09
Code: https://github.com/HalfVulpes/sts2ai
Terminology
Sources
- Deep Reinforcement Learning at the Edge of the Statistical Precipice
- AgenticSTS: A Bounded-Memory Testbed for Long-Horizon LLM Agents
- Per-Context Reverse Curricula for Sparse-Reward Reinforcement Learning
- First return, then explore
- Reverse Curriculum Generation for Reinforcement Learning
- Prioritized Level Replay
- Empirical Design in Reinforcement Learning
- Backplay: "Man muss immer umkehren"
- Proximal Policy Optimization Algorithms
- SpireBench: Evaluating Long-Horizon Strategic Play of Agentic Models with Slay the Spire
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis