Reinforce Agent playing Pixelcopter-PLE-v0
This is a trained REINFORCE policy for PixelCopter from Unit 4 of the Hugging Face Deep RL course.
Best checkpoint: h64_lr1e-4_seed50 at episode 9000.
Replay video uses greedy actions from the learned policy.
Evaluation
- Greedy mean_reward: 34.90
- Greedy std_reward: 15.33
- Greedy score mean-std: 19.57
- Sample mean_reward: 32.20
- Sample std_reward: 20.54
- Sample score mean-std: 11.66
Evaluation results
- mean_reward on Pixelcopter-PLE-v0self-reported34.90 +/- 15.33