- Implemented policy_utils.py with helper functions for action selection, including epsilon-greedy support.

- Updated `requirements.txt` to relax PyTorch version constraint for better GPU compatibility.
- Added detailed GPU setup instructions, new device fallback options, and command examples to `README.md`.
- Developed a new script `plot_model_max_x_trend.py` for visualizing training trends, generating HTML/Markdown reports.

mario-rl-mvp/requirements.txt

@@ -1,4 +1,5 @@
-torch==2.5.1
+# Keep torch unpinned to avoid forcing old wheels on new GPUs (e.g. RTX 50xx).
+torch>=2.5.1
 stable-baselines3==2.3.2
 gym==0.26.2
 gymnasium==0.29.1