Reinforcement Learning on CartPole Gymnasium
500Mean Best RewardProPS+
Evaluation results
| Method | Links | |
|---|---|---|
| ProPS+Episodes per run=8,000, Language Model backbone=GPT-4o, Independent runs=102026.05 | 500 | |
| R2POEpisodes per run=4,000, Language Model backbone=gpt-oss:20b, Independent runs=102026.05 | 500 | |
| ProPSEpisodes per run=8,000, Language Model backbone=GPT-4o, Independent runs=102026.05 | 478.27 | |
| TRPOEpisodes per run=8,000, Independent runs=102026.05 | 465.34 | |
| PPOEpisodes per run=8,000, Independent runs=102026.05 | 365.86 |