lnp_ml/models/pretrain/llm/pretrain_history.json

{
  "train": [
    {
      "loss": 0.8448960826835799,
      "n_samples": 8236
    },
    {
      "loss": 0.719834098745284,
      "n_samples": 8236
    },
    {
      "loss": 0.663177823979858,
      "n_samples": 8236
    },
    {
      "loss": 0.6313324196064223,
      "n_samples": 8236
    },
    {
      "loss": 0.6015973463980187,
      "n_samples": 8236
    },
    {
      "loss": 0.5772233604516376,
      "n_samples": 8236
    },
    {
      "loss": 0.5503088645184717,
      "n_samples": 8236
    },
    {
      "loss": 0.5416200838950492,
      "n_samples": 8236
    },
    {
      "loss": 0.5272663477672541,
      "n_samples": 8236
    },
    {
      "loss": 0.5049150672104823,
      "n_samples": 8236
    },
    {
      "loss": 0.4867241953315522,
      "n_samples": 8236
    },
    {
      "loss": 0.48287810899784056,
      "n_samples": 8236
    },
    {
      "loss": 0.46680330384871865,
      "n_samples": 8236
    },
    {
      "loss": 0.4364163791050663,
      "n_samples": 8236
    },
    {
      "loss": 0.41805977998513977,
      "n_samples": 8236
    },
    {
      "loss": 0.41134080713440224,
      "n_samples": 8236
    },
    {
      "loss": 0.40551714557881147,
      "n_samples": 8236
    },
    {
      "loss": 0.39040500128251826,
      "n_samples": 8236
    },
    {
      "loss": 0.39093497016292106,
      "n_samples": 8236
    },
    {
      "loss": 0.38722379066343154,
      "n_samples": 8236
    },
    {
      "loss": 0.36881573111064925,
      "n_samples": 8236
    },
    {
      "loss": 0.3569791747592733,
      "n_samples": 8236
    },
    {
      "loss": 0.35541044713340375,
      "n_samples": 8236
    },
    {
      "loss": 0.3500091104527131,
      "n_samples": 8236
    }
  ],
  "val": [
    {
      "loss": 0.8445101732073,
      "n_samples": 1454
    },
    {
      "loss": 1.7723233459576289,
      "n_samples": 1454
    },
    {
      "loss": 0.7529299004205812,
      "n_samples": 1454
    },
    {
      "loss": 1.0581415233290672,
      "n_samples": 1454
    },
    {
      "loss": 1.0012581408269796,
      "n_samples": 1454
    },
    {
      "loss": 0.7138602705736435,
      "n_samples": 1454
    },
    {
      "loss": 0.658381061917978,
      "n_samples": 1454
    },
    {
      "loss": 0.7822830723765136,
      "n_samples": 1454
    },
    {
      "loss": 0.7124623070884768,
      "n_samples": 1454
    },
    {
      "loss": 0.6592001183652813,
      "n_samples": 1454
    },
    {
      "loss": 1.0880131611633825,
      "n_samples": 1454
    },
    {
      "loss": 0.7002065947491988,
      "n_samples": 1454
    },
    {
      "loss": 0.6979960435194359,
      "n_samples": 1454
    },
    {
      "loss": 0.6506489415280264,
      "n_samples": 1454
    },
    {
      "loss": 0.7450973976101147,
      "n_samples": 1454
    },
    {
      "loss": 0.7587803819484527,
      "n_samples": 1454
    },
    {
      "loss": 0.734511285159414,
      "n_samples": 1454
    },
    {
      "loss": 0.8079460347699003,
      "n_samples": 1454
    },
    {
      "loss": 0.7549331074239791,
      "n_samples": 1454
    },
    {
      "loss": 0.7144972918942017,
      "n_samples": 1454
    },
    {
      "loss": 0.6926865217774245,
      "n_samples": 1454
    },
    {
      "loss": 0.6760478137775825,
      "n_samples": 1454
    },
    {
      "loss": 0.6954523239699814,
      "n_samples": 1454
    },
    {
      "loss": 0.6800617901134359,
      "n_samples": 1454
    }
  ]
}