{
  "benchmarked_devices": [
    "cuda"
  ],
  "parameters": {
    "total_params": 27833970,
    "trainable_params": 14304786,
    "total_params_m": 27.83397,
    "trainable_params_m": 14.304786
  },
  "cuda": {
    "memory": {
      "baseline_memory_mb": 288.01806640625,
      "peak_memory_mb": 302.52978515625,
      "current_memory_mb": 288.5927734375,
      "inference_memory_mb": 14.51171875,
      "device": "cuda"
    },
    "inference_single": {
      "batch_size": 1,
      "mean_latency_ms": 3.333154039864894,
      "std_latency_ms": 0.017885221990080365,
      "median_latency_ms": 3.3323254974675365,
      "p95_latency_ms": 3.342447954128147,
      "p99_latency_ms": 3.4173157204349995,
      "min_latency_ms": 3.309674000774976,
      "max_latency_ms": 3.4711439948296174,
      "latency_per_image_ms": 3.333154039864894,
      "throughput_img_per_sec": 300.01613728015224,
      "device": "cuda"
    },
    "inference_batch32": {
      "batch_size": 32,
      "mean_latency_ms": 43.753516640135786,
      "std_latency_ms": 0.14937650685647402,
      "median_latency_ms": 43.77655049756868,
      "p95_latency_ms": 43.96002150133427,
      "p99_latency_ms": 44.032208782446105,
      "min_latency_ms": 43.447225994896144,
      "max_latency_ms": 44.07967900624499,
      "latency_per_image_ms": 1.3672973950042433,
      "throughput_img_per_sec": 731.3697836723345,
      "device": "cuda"
    }
  }
}