{
  "elapsed_seconds": 1349.062306778971,
  "evaluated_at_unix": 1793127069.5127065,
  "held_out_tokens": 101191373,
  "model": "monody",
  "objective": "next-token NLL",
  "optimizer_step": 2352,
  "pass": 7,
  "stage": "monody-pass-7",
  "training_tokens": 616562688,
  "unique_training_tokens": 88080384,
  "weights": {
    "ema": {
      "downstream": {
        "arc_easy": {
          "accuracy": 0.475,
          "accuracy_norm": 0.433,
          "chance": 0.25043333333333334,
          "elapsed_seconds": 7.513357853982598,
          "examples": 1000,
          "gold_bits_per_byte": 1.2626438911998918
        },
        "hellaswag": {
          "accuracy": 0.264,
          "accuracy_norm": 0.279,
          "chance": 0.25,
          "elapsed_seconds": 8.59502274915576,
          "examples": 1000,
          "gold_bits_per_byte": 1.1487319894695152
        },
        "lambada": {
          "accuracy": 0.217,
          "accuracy_norm": null,
          "chance": 0.0,
          "elapsed_seconds": 3.444128228817135,
          "examples": 1000,
          "gold_bits_per_byte": 1.2708033027176069,
          "perplexity": 355.8159137348601
        },
        "mean_accuracy": 0.45649999999999996,
        "mean_gold_bits_per_byte": 1.3049644023078721,
        "piqa": {
          "accuracy": 0.564,
          "accuracy_norm": 0.564,
          "chance": 0.5,
          "elapsed_seconds": 5.2499302718788385,
          "examples": 1000,
          "gold_bits_per_byte": 1.3786440568762732
        },
        "sciq": {
          "accuracy": 0.726,
          "accuracy_norm": 0.636,
          "chance": 0.25,
          "elapsed_seconds": 8.13590192887932,
          "examples": 1000,
          "gold_bits_per_byte": 1.408172402357161
        },
        "winogrande": {
          "accuracy": 0.493,
          "accuracy_norm": null,
          "chance": 0.5,
          "elapsed_seconds": 4.800854941364378,
          "examples": 1000,
          "gold_bits_per_byte": 1.3607907712267844
        }
      },
      "generalization_gap": 1.5875091574716764,
      "held_out": {
        "elapsed_seconds": 534.5517211458646,
        "loss": 3.6317696267661397,
        "loss_95ci": [
          3.6271775943594977,
          3.636572309692387
        ],
        "objective": "next-token NLL",
        "perplexity": 37.77961331207202,
        "tokens": 101191373,
        "tokens_per_second": 189301.3697216918,
        "windows": 49410
      },
      "memorization": {
        "held_out": {
          "continuation_tokens": 32,
          "exact_match": 0.0,
          "prefix_tokens": 64,
          "token_accuracy": 0.0364990234375,
          "windows": 256
        },
        "train": {
          "continuation_tokens": 32,
          "exact_match": 0.00390625,
          "prefix_tokens": 64,
          "token_accuracy": 0.0614013671875,
          "windows": 256
        }
      },
      "memorization_gap": 0.00390625,
      "train_probe": {
        "elapsed_seconds": 54.62738667288795,
        "loss": 2.0442604692944633,
        "loss_95ci": [
          2.036346034331813,
          2.052173914243236
        ],
        "objective": "next-token NLL",
        "perplexity": 7.72344470088251,
        "tokens": 10000384,
        "tokens_per_second": 183065.39281995854,
        "windows": 4883
      }
    },
    "normal": {
      "downstream": null,
      "generalization_gap": 1.5069047089964025,
      "held_out": {
        "elapsed_seconds": 637.9913326767273,
        "loss": 3.740345572211005,
        "loss_95ci": [
          3.7358146551118874,
          3.745179232388174
        ],
        "objective": "next-token NLL",
        "perplexity": 42.11254057450087,
        "tokens": 101191373,
        "tokens_per_second": 158609.32244243208,
        "windows": 49410
      },
      "memorization": {
        "held_out": {
          "continuation_tokens": 32,
          "exact_match": 0.0,
          "prefix_tokens": 64,
          "token_accuracy": 0.0335693359375,
          "windows": 256
        },
        "train": {
          "continuation_tokens": 32,
          "exact_match": 0.00390625,
          "prefix_tokens": 64,
          "token_accuracy": 0.0570068359375,
          "windows": 256
        }
      },
      "memorization_gap": 0.00390625,
      "train_probe": {
        "elapsed_seconds": 67.42786940094084,
        "loss": 2.2334408632146023,
        "loss_95ci": [
          2.223836386290301,
          2.2425792741355606
        ],
        "objective": "next-token NLL",
        "perplexity": 9.331920762841568,
        "tokens": 10000384,
        "tokens_per_second": 148312.32380390863,
        "windows": 4883
      }
    }
  }
}