Reinforcement Learning on Skiing simplified environment
5,492Average Episodic ReturnNeuralPPO
Evaluation Results
| Method | Links | |
|---|---|---|
| NeuralPPO2026.04 | 5,492 | |
| BlendRL+ClaudeValuation source=Claude2026.04 | 5,388 | |
| GRAIL (GPT-4o)Valuation source=GPT-4o, Concept-alignment coefficient (cCA)=0.12026.04 | 5,253 | |
| BlendRL+GPT-4oValuation source=GPT-4o2026.04 | 5,171 | |
| BlendRL (no CA)Concept Alignment (CA)=false, Concept-alignment coefficient (cCA)=02026.04 | 5,086 | |
| GRAIL (Claude)Valuation source=Claude, Concept-alignment coefficient (cCA)=0.12026.04 | 5,021 |