Reinforcement Learning on Pong Gymnasium
3Mean Best RewardR2PO
Evaluation results
| Method | Links | |
|---|---|---|
| R2POEpisodes per run=4,000, Language Model backbone=gpt-oss:20b, Independent runs=102026.05 | 3 | |
| ProPS+Episodes per run=8,000, Language Model backbone=GPT-4o, Independent runs=102026.05 | 2.99 | |
| ProPSEpisodes per run=8,000, Language Model backbone=GPT-4o, Independent runs=102026.05 | 2.8 | |
| PPOEpisodes per run=8,000, Independent runs=102026.05 | 2.29 | |
| TRPOEpisodes per run=8,000, Independent runs=102026.05 | 1.36 |