benchmark_results.json

{
  "model_params": 24384000,
  "total_correct": 40,
  "total_tests": 72,
  "overall_pct": 55.55555555555556,
  "elapsed_s": 81.88592767715454,
  "categories": {
    "1. Addition (basic)": {
      "correct": 6,
      "total": 10,
      "pct": 60.0
    },
    "2. Subtraction": {
      "correct": 4,
      "total": 10,
      "pct": 40.0
    },
    "3. Multiplication": {
      "correct": 6,
      "total": 10,
      "pct": 60.0
    },
    "4. Division": {
      "correct": 9,
      "total": 10,
      "pct": 90.0
    },
    "5. Multi-step": {
      "correct": 3,
      "total": 6,
      "pct": 50.0
    },
    "6. Algebra (linear)": {
      "correct": 6,
      "total": 8,
      "pct": 75.0
    },
    "7. Sequences": {
      "correct": 1,
      "total": 5,
      "pct": 20.0
    },
    "8. Russian language": {
      "correct": 0,
      "total": 5,
      "pct": 0.0
    },
    "9. Knowledge": {
      "correct": 4,
      "total": 5,
      "pct": 80.0
    },
    "10. Word problems": {
      "correct": 1,
      "total": 3,
      "pct": 33.333333333333336
    }
  }
}