Adding Hardware-Specific Configs and Evaluation Results for RL Training

[ravenouse]

I believe it would be great to add more config YAML files with hardware details (e.g., GPU type, memory) and, if possible, evaluation results of trained models. This would make RL training more accessible and easier to replicate for users with diverse setups.