{
  "experiment": "tiny trained causal language-model baseline",
  "honesty": "Educational synthetic grammar experiment only; it is not DeepSeek and does not establish natural-language generalization.",
  "device": "cpu",
  "torch_version": "2.8.0",
  "config": {
    "dataset_seed": 20260920,
    "split_seed": 314159,
    "training_seed": 20260920,
    "train_fraction": 0.8,
    "layers": 2,
    "dim": 48,
    "heads": 4,
    "mlp_dim": 96,
    "context": 24,
    "batch_size": 32,
    "steps": 600,
    "learning_rate": 0.003,
    "weight_decay": 0.0,
    "eval_every": 100,
    "torch_threads": 2
  },
  "parameters": 41088,
  "vocab_size": 20,
  "train_sentence_count": 345,
  "validation_sentence_count": 87,
  "validation_target_tokens": 957,
  "metrics": {
    "uniform_validation_ce": 2.995732273553991,
    "smoothed_train_unigram_validation_ce": 2.763847827911377,
    "untrained_validation_ce": 3.370097990444586,
    "trained_validation_ce": 0.5721415935265233,
    "curve": [
      {
        "step": 0,
        "train_ce": 3.379961281677166,
        "validation_ce": 3.370097990444586
      },
      {
        "step": 100,
        "train_ce": 0.5631159605244874,
        "validation_ce": 0.5700990045083105
      },
      {
        "step": 200,
        "train_ce": 0.5610740691776803,
        "validation_ce": 0.5727658645486383
      },
      {
        "step": 300,
        "train_ce": 0.5597752627647913,
        "validation_ce": 0.5789237824602436
      },
      {
        "step": 400,
        "train_ce": 0.557131049315763,
        "validation_ce": 0.5773473119785544
      },
      {
        "step": 500,
        "train_ce": 0.5541823476356637,
        "validation_ce": 0.5650235996226407
      },
      {
        "step": 600,
        "train_ce": 0.5546246957087863,
        "validation_ce": 0.5721415935265233
      }
    ]
  },
  "causal_mask_invariance": {
    "max_abs_difference_at_earlier_positions": 0.0,
    "passed": true
  },
  "fixed_prompt_samples": {
    "prompt": "<bos> move the red circle",
    "before": "<bos> move the red circle the the <bos> green it twice the <bos>",
    "after": "<bos> move the red circle down once then paint it green <eos>"
  },
  "runtime_seconds": 1.7079257500590757,
  "checkpoint": "trained.pt"
}