plots/training_summary.json

{
  "model": "qwen2.5:3b-instruct-q4_K_M",
  "device": "M4 Mac (Ollama local)",
  "training_rounds": 4,
  "episodes_per_round": 6,
  "before": {
    "monthly_engage": 0.3548,
    "monthly_strategic": 0.6795,
    "monthly_competitive": 0.3738
  },
  "after": {
    "monthly_engage": 0.4086,
    "monthly_strategic": 0.6273,
    "monthly_competitive": 0.5101
  },
  "smart_heuristic": {
    "monthly_engage": 0.4312,
    "monthly_strategic": 0.7682,
    "monthly_competitive": 0.8094
  },
  "improvement": {
    "monthly_engage": 0.053800000000000014,
    "monthly_strategic": -0.052200000000000024,
    "monthly_competitive": 0.13629999999999998
  },
  "training_log": {
    "round": [
      1,
      2,
      3,
      4
    ],
    "avg_grader": [
      0.4958,
      0.4912,
      0.6015,
      0.5548
    ],
    "max_grader": [
      0.7391,
      0.7236,
      0.7529,
      0.7705
    ],
    "min_grader": [
      0.3698,
      0.2527,
      0.382,
      0.3764
    ],
    "avg_reward": [
      6.07,
      6.093,
      6.418,
      6.467
    ],
    "max_reward": [
      6.104,
      6.1,
      6.481,
      6.527
    ],
    "min_reward": [
      6.037,
      6.076,
      6.343,
      6.366
    ],
    "best_temperature": [
      1.4,
      1.0,
      0.7,
      0.7
    ]
  },
  "all_episodes": [
    {
      "round": 1,
      "task": "monthly_engage",
      "seed": 42,
      "grader_score": 0.4395,
      "total_reward": 6.1044,
      "temperature": 1.4
    },
    {
      "round": 1,
      "task": "monthly_strategic",
      "seed": 43,
      "grader_score": 0.6758,
      "total_reward": 6.0373,
      "temperature": 1.4
    },
    {
      "round": 1,
      "task": "monthly_competitive",
      "seed": 44,
      "grader_score": 0.3698,
      "total_reward": 6.0686,
      "temperature": 1.4
    },
    {
      "round": 1,
      "task": "monthly_engage",
      "seed": 45,
      "grader_score": 0.3806,
      "total_reward": 6.0643,
      "temperature": 1.4
    },
    {
      "round": 1,
      "task": "monthly_strategic",
      "seed": 46,
      "grader_score": 0.7391,
      "total_reward": 6.096,
      "temperature": 1.4
    },
    {
      "round": 1,
      "task": "monthly_competitive",
      "seed": 47,
      "grader_score": 0.3699,
      "total_reward": 6.0489999999999995,
      "temperature": 1.4
    },
    {
      "round": 2,
      "task": "monthly_engage",
      "seed": 142,
      "grader_score": 0.4335,
      "total_reward": 6.0995,
      "temperature": 1.0
    },
    {
      "round": 2,
      "task": "monthly_strategic",
      "seed": 143,
      "grader_score": 0.7236,
      "total_reward": 6.0992,
      "temperature": 1.0
    },
    {
      "round": 2,
      "task": "monthly_competitive",
      "seed": 144,
      "grader_score": 0.3789,
      "total_reward": 6.0943,
      "temperature": 1.0
    },
    {
      "round": 2,
      "task": "monthly_engage",
      "seed": 145,
      "grader_score": 0.4356,
      "total_reward": 6.0999,
      "temperature": 1.0
    },
    {
      "round": 2,
      "task": "monthly_strategic",
      "seed": 146,
      "grader_score": 0.7232,
      "total_reward": 6.0882,
      "temperature": 1.0
    },
    {
      "round": 2,
      "task": "monthly_competitive",
      "seed": 147,
      "grader_score": 0.2527,
      "total_reward": 6.0764,
      "temperature": 1.0
    },
    {
      "round": 3,
      "task": "monthly_engage",
      "seed": 242,
      "grader_score": 0.382,
      "total_reward": 6.4364,
      "temperature": 0.7
    },
    {
      "round": 3,
      "task": "monthly_strategic",
      "seed": 243,
      "grader_score": 0.6426,
      "total_reward": 6.4364,
      "temperature": 0.7
    },
    {
      "round": 3,
      "task": "monthly_competitive",
      "seed": 244,
      "grader_score": 0.7529,
      "total_reward": 6.3849,
      "temperature": 0.7
    },
    {
      "round": 3,
      "task": "monthly_engage",
      "seed": 245,
      "grader_score": 0.3935,
      "total_reward": 6.4805,
      "temperature": 0.7
    },
    {
      "round": 3,
      "task": "monthly_strategic",
      "seed": 246,
      "grader_score": 0.724,
      "total_reward": 6.4286,
      "temperature": 0.7
    },
    {
      "round": 3,
      "task": "monthly_competitive",
      "seed": 247,
      "grader_score": 0.7138,
      "total_reward": 6.3425,
      "temperature": 0.7
    },
    {
      "round": 4,
      "task": "monthly_engage",
      "seed": 342,
      "grader_score": 0.3764,
      "total_reward": 6.4858,
      "temperature": 0.7
    },
    {
      "round": 4,
      "task": "monthly_strategic",
      "seed": 343,
      "grader_score": 0.6314,
      "total_reward": 6.4636,
      "temperature": 0.7
    },
    {
      "round": 4,
      "task": "monthly_competitive",
      "seed": 344,
      "grader_score": 0.7705,
      "total_reward": 6.4934,
      "temperature": 0.7
    },
    {
      "round": 4,
      "task": "monthly_engage",
      "seed": 345,
      "grader_score": 0.3851,
      "total_reward": 6.4661,
      "temperature": 0.7
    },
    {
      "round": 4,
      "task": "monthly_strategic",
      "seed": 346,
      "grader_score": 0.6755,
      "total_reward": 6.5269,
      "temperature": 0.7
    },
    {
      "round": 4,
      "task": "monthly_competitive",
      "seed": 347,
      "grader_score": 0.4897,
      "total_reward": 6.3657,
      "temperature": 0.7
    }
  ],
  "elapsed_seconds": 6034.9
}