README.md

metadata

tags:
  - reinforcement-learning
  - pytorch
  - custom-implementation
  - MountainCarContinuous-v0
pipeline_tag: reinforcement-learning
library_name: PyTorch
model-index:
  - name: FDF-MountainCarContinuous-v0
    results:
      - task:
          type: reinforcement-learning
          name: reinforcement-learning
        dataset:
          name: MountainCarContinuous-v0
          type: MountainCarContinuous-v0
        metrics:
          - type: mean_reward
            value: 97.1
            name: mean_reward
          - type: std_reward
            value: 0.65
            name: std_reward

FDF Canopy: MountainCarContinuous-v0 (Score: 97.10)

This repository contains the frozen PyTorch model weights for a custom architecture that shattered the theoretical physics ceiling of MountainCarContinuous-v0.

Methodology

Instead of Reward Shaping or augmented logic, this agent utilizes an unmodified PPO core wrapped in an FDF governor. The governor injects a massive curiosity penalty (Beta 5.0) based purely on neural topological stagnation during the Play Phase, seamlessly decaying to 0.0 to allow raw physics compilation during the Mastery Phase.

• Governor Beta: 5.0
• Schedule: 50% Play / 30% Decay
• Hardware Bias: Apple Silicon (MPS)
• Holdout Average (100-Episodes): 97.10

How to Evaluate

1. Download the fdf_seed_24_golden_mountaincar.pt file.
2. Download the fdf_inference.py script to instantiate the custom PyTorch ContinuousPPOMLP.
3. Run the evaluation natively using:

   python fdf_inference.py --model_path fdf_seed_24_golden_mountaincar.pt