PPO Atari Pong RAM Artifacts

This file visualizes the training curve produced by NN/Examples/Models/RL/PPOPongRam.lean (lake exe torchlean ppo_pong_ram).

Pong RAM intentionally uses the same Gymnasium boundary as CartPole, but with ALE registration and a higher-dimensional observation. It is an integration/regression example, not a tuned Atari agent.

Workflow:

1. Run:

python3 -m pip install --user 'gymnasium>=1.0' ale-py
lake exe torchlean ppo_pong_ram
lake build -R -K cuda=true && lake exe torchlean ppo_pong_ram --cuda

2. Put the cursor on the command below in an editor. The infoview will render the saved log.

Notes:

• The executable writes data/rl/ppo_pong_ram_trainlog.json by default (override with --log).
• This viewer is pure: if the file is missing, it shows an error panel instead of failing to build.

References:

• Schulman et al., "Proximal Policy Optimization Algorithms" (2017): https://arxiv.org/abs/1707.06347
• Machado et al., "Revisiting the Arcade Learning Environment" (2018): https://arxiv.org/abs/1709.06009
• ALE docs: https://ale.farama.org/