{
  "parameter_count": 2767529,
  "training_steps": 28110,
  "epochs_completed": 30,
  "final_recent_loss": 0.01534514181315899,
  "training_seconds": 4215.443364993001,
  "args": {
    "output_dir": "runs/tiny_digit_diffusion_3m",
    "data_dir": "/home/codex/data/mnist",
    "epochs": 30,
    "samples_per_epoch": 60000,
    "batch_size": 64,
    "learning_rate": 0.0002,
    "weight_decay": 0.0,
    "warmup_steps": 500,
    "grad_clip": 1.0,
    "condition_dropout": 0.1,
    "ema_decay": 0.999,
    "num_workers": 4,
    "max_steps": 0,
    "log_steps": 50,
    "sample_every": 1,
    "sample_steps": 40,
    "guidance_scale": 3.0,
    "checkpoint_every": 5,
    "seed": 1234,
    "device": "cuda"
  },
  "config": {
    "image_height": 32,
    "image_width": 256,
    "max_digits": 8,
    "slot_width": 32,
    "base_channels": 48,
    "channel_mults": [
      1,
      1.5,
      2,
      2.6666666666666665
    ],
    "embedding_dim": 192,
    "token_embedding_dim": 32,
    "condition_channels": 8,
    "attention_heads": 4,
    "diffusion_steps": 400
  },
  "sample_prompts": [
    "0",
    "7",
    "42",
    "2026",
    "12345678",
    "99999999"
  ],
  "recommended_inference": {
    "sampling_steps": 50,
    "guidance_scale": 1.0
  },
  "evaluation": {
    "classifier_test_accuracy": 0.9868,
    "generated_digit_accuracy": 0.926,
    "generated_sequence_exact_match": 0.725,
    "evaluation_prompts": 240,
    "prompts_per_length_1_to_8": 30
  }
}