training_log.json

[
  {
    "epoch": 1,
    "total_loss": 3.2776,
    "ce_loss": 2.4249,
    "kl_loss": 4.1302,
    "lr": 0.0009931853456062877,
    "time_seconds": 473.7
  },
  {
    "epoch": 2,
    "total_loss": 1.0516,
    "ce_loss": 1.3657,
    "kl_loss": 0.7376,
    "lr": 0.0009397775952968608,
    "time_seconds": 472.9
  },
  {
    "epoch": 3,
    "total_loss": 0.7804,
    "ce_loss": 1.1816,
    "kl_loss": 0.3793,
    "lr": 0.0008387456967629545,
    "time_seconds": 472.5
  },
  {
    "epoch": 4,
    "total_loss": 0.6937,
    "ce_loss": 1.1151,
    "kl_loss": 0.2724,
    "lr": 0.0007010305395874761,
    "time_seconds": 472.8
  },
  {
    "epoch": 5,
    "total_loss": 0.6475,
    "ce_loss": 1.0745,
    "kl_loss": 0.2205,
    "lr": 0.0005415454961846232,
    "time_seconds": 472.6
  },
  {
    "epoch": 6,
    "total_loss": 0.6168,
    "ce_loss": 1.0443,
    "kl_loss": 0.1892,
    "lr": 0.00037756143125471743,
    "time_seconds": 472.3
  },
  {
    "epoch": 7,
    "total_loss": 0.5937,
    "ce_loss": 1.019,
    "kl_loss": 0.1684,
    "lr": 0.00022683641500914116,
    "time_seconds": 472.9
  },
  {
    "epoch": 8,
    "total_loss": 0.5759,
    "ce_loss": 0.9977,
    "kl_loss": 0.1542,
    "lr": 0.0001056926762098463,
    "time_seconds": 472.8
  },
  {
    "epoch": 9,
    "total_loss": 0.5637,
    "ce_loss": 0.9817,
    "kl_loss": 0.1456,
    "lr": 2.724904454425836e-05,
    "time_seconds": 472.9
  },
  {
    "epoch": 10,
    "total_loss": 0.5573,
    "ce_loss": 0.9727,
    "kl_loss": 0.1419,
    "lr": 2.936926663532624e-10,
    "time_seconds": 472.3
  }
]