# Эксперимент: Backward-Reachability Distance Head (#1375)

{
  "worked": false,
  "confidence": 8,
  "verdict": "Built a small actor-critic MVP with an exact backward-reachability distance table, Huber distance head, and modal progress hinge loss in experiment.py. The numerical math check passed: distances matched shortest paths, greedy transitions had zero progress violations, and the ideal progress hinge was zero. In the matched two-seed toy experiment, the auxiliary method slightly improved late training reward (0.750 vs 0.713) but did not improve evaluation success (0.500 vs 0.500) or steps to success (4.0 vs 4.0), so the claimed learning benefit was not clearly observed.",
  "metrics": {
    "baseline": "Late training reward 0.7125; evaluation success 0.500; successful-episode steps 4.0; critic loss 0.07270.",
    "idea": "Late training reward 0.750; evaluation success 0.500; successful-episode steps 4.0; critic loss 0.07075; distance-head absolute error 2.746."
  },
  "how_to_run": "/home/maxwelhelp/main/bin/python3 experiment.py",
  "files": [
    "experiment.py",
    "results.json"
  ],
  "limitations": "Only a fully observable nine-state chain was tested, with two random seeds and 80 updates per condition. No PPO clipping, MiniGrid layouts, hazards, partial observability/GRU head, held-out layouts, or statistically powered multi-seed evaluation was tested."
}