quantization_report.json

{
  "original": {
    "auroc": 0.8064,
    "sensitivity": 0.7489,
    "specificity": 0.6954,
    "accuracy": 0.7048,
    "avg_inference_ms": 2.061,
    "size_mb": 3.8
  },
  "quantized": {
    "auroc": 0.6956,
    "sensitivity": 0.8052,
    "specificity": 0.4806,
    "accuracy": 0.5378,
    "avg_inference_ms": 0.484,
    "size_mb": 1.47
  },
  "compression": {
    "original_size_mb": 3.8,
    "quantized_size_mb": 1.47,
    "size_reduction_pct": 61.3,
    "size_ratio": 2.59
  },
  "auroc_drop": 0.1108,
  "created": "2026-04-23T17:16:41.961615"
}