{
  "parameters": 28480,
  "config": {
    "vocab_size": 27,
    "context_length": 48,
    "embed_dim": 32,
    "num_heads": 4,
    "num_layers": 2
  },
  "seed": 17,
  "steps": 500,
  "session_steps": 500,
  "device": "cpu",
  "training_seconds": 1.8433433340396732,
  "stop_reason": "step_limit",
  "before": {
    "train_loss": 3.288892698287964,
    "validation_loss": 3.289594268798828,
    "sample": "the mvnxsfpmdxry\nevynphwyhvwgdf qkc vdreeaiyeteadtumkmiahiwoi.wiucyidshptrg.cdaiupem"
  },
  "after": {
    "train_loss": 0.22508766651153564,
    "validation_loss": 0.23227010071277618,
    "sample": "the hill.\nthe small dog rests near the hill.\nthe quiet fox sleeps near the gardi the"
  },
  "history": [
    {
      "step": 0,
      "train_loss": 3.288892698287964,
      "validation_loss": 3.289594268798828
    },
    {
      "step": 50,
      "train_loss": 1.3840496778488158,
      "validation_loss": 1.3859103679656983
    },
    {
      "step": 100,
      "train_loss": 0.8640612125396728,
      "validation_loss": 0.8863992929458618
    },
    {
      "step": 150,
      "train_loss": 0.5367879092693328,
      "validation_loss": 0.5650198221206665
    },
    {
      "step": 200,
      "train_loss": 0.3869510889053345,
      "validation_loss": 0.4091724455356598
    },
    {
      "step": 250,
      "train_loss": 0.32893029451370237,
      "validation_loss": 0.3371934473514557
    },
    {
      "step": 300,
      "train_loss": 0.2845767676830292,
      "validation_loss": 0.2938975811004639
    },
    {
      "step": 350,
      "train_loss": 0.27053363621234894,
      "validation_loss": 0.2817482233047485
    },
    {
      "step": 400,
      "train_loss": 0.2578303664922714,
      "validation_loss": 0.27535892426967623
    },
    {
      "step": 450,
      "train_loss": 0.2475225269794464,
      "validation_loss": 0.2603019118309021
    },
    {
      "step": 500,
      "train_loss": 0.22508766651153564,
      "validation_loss": 0.23227010071277618
    }
  ],
  "tests": {
    "checkpoint_reload_exact": true,
    "causal_prefix_unchanged": true,
    "held_out_loss_improved": true
  },
  "dataset": {
    "description": "Original synthetic nature sentences, split before window sampling.",
    "train_documents": 316,
    "validation_documents": 92,
    "sha256": "abe7f4f04ab6b0d4c198b67f4f738fa2798a1eac4e676a1464f818ffd28f11e3",
    "limitations": "Shared grammar and vocabulary. Not evidence of broad language understanding."
  }
}