{
  "environment": "4x4 grid; start(0,0); goal(3,3); obstacles(1,1),(2,1),(1,3)",
  "algorithm": "tabular Q-learning",
  "seed": 17,
  "episodes": 80,
  "stepCap": 40,
  "alpha": 0.5,
  "gamma": 0.9,
  "epsilon": 0.25,
  "reward": "-1/move; -4 extra collision; +20 extra goal; terminal has no bootstrap",
  "provenance": "Generated by src/data/program-labs.ts; reproduced by the Python exercise",
  "returns": [
    -6, -40, -48, -23, 5, 14, 7, 4, -13, 4, 12, 12, 1, 2, 14, 12, -1, 14, -1, 9,
    4, 14, 12, 14, 12, 14, 12, 9, 14, 14, 4, 14, 9, 2, 14, 14, 7, 4, 14, 2, 5,
    14, 14, 6, 10, 4, 9, 14, 14, 2, 14, 14, 12, 2, 14, 9, 14, 2, 9, 8, 4, 2, 14,
    12, 12, 14, 7, 4, 4, 9, 14, 9, 14, 9, 10, 14, 2, 14, 4, -1
  ]
}