{
  "1": {
    "rows": 50,
    "first_episode": 2,
    "last_episode": 100,
    "reward_first": -0.32,
    "reward_last": -0.21,
    "reward_slope_per_episode": 0.001393037214885956,
    "reward_rolling_first": -0.36000000000000004,
    "reward_rolling_last": -0.276,
    "reward_rolling_max": -0.11000000000000001,
    "reward_rolling_max_episode": 44,
    "sent_first": 35.3356890459364,
    "sent_last": 33.57142857142857,
    "validation": [
      {
        "episode": 50,
        "episode_reward": -0.25123363949665195,
        "episode_reward_std": 0.16381281200573786,
        "sent_percentage": 37.74333002499273
      },
      {
        "episode": 100,
        "episode_reward": -0.1813911549411824,
        "episode_reward_std": 0.16822846257367927,
        "sent_percentage": 41.23330359876366
      }
    ]
  },
  "2": {
    "rows": 49,
    "first_episode": 2,
    "last_episode": 98,
    "reward_first": -0.41,
    "reward_last": -0.18,
    "reward_slope_per_episode": 0.002081632653061224,
    "reward_rolling_first": -0.43000000000000005,
    "reward_rolling_last": -0.156,
    "reward_rolling_max": -0.156,
    "reward_rolling_max_episode": 98,
    "sent_first": 30.508474576271187,
    "sent_last": 30.097087378640776,
    "validation": [
      {
        "episode": 50,
        "episode_reward": -0.2394787649321464,
        "episode_reward_std": 0.17085465251533977,
        "sent_percentage": 38.33082568972942
      },
      {
        "episode": 100,
        "episode_reward": -0.2106141229773678,
        "episode_reward_std": 0.1559543060504657,
        "sent_percentage": 39.77294244656109
      }
    ]
  },
  "scope": "training points are the mean of two rollout threads, logged every two environment episodes; rolling lines use five logged points (ten environment episodes); validation points are 32-city summaries"
}