benchmark/benchmark.json

{
  "parameters": 1913971,
  "counted_parameters": 1913971,
  "macs": 2572161600,
  "gmacs": 2.5721616,
  "flops_estimate": 5144323200,
  "gflops_estimate": 5.1443232,
  "mean_latency_ms": 11.716616345202056,
  "median_latency_ms": 11.059291500714608,
  "p95_latency_ms": 16.591671999776736,
  "fps": 85.34887296275593,
  "model": "outputs/finetune_direct/p20_adamw_exact/weights/best.pt",
  "model_size_mb": 3.8965396881103516,
  "batch_size": 1,
  "imgsz": 640,
  "device": "cuda:1",
  "gpu_name": "Tesla T4",
  "gpu_total_memory_mb": 14911.6875,
  "peak_gpu_memory_mb": 34.1357421875,
  "python_version": "3.12.12",
  "torch_version": "2.10.0+cu128",
  "cuda_version": "12.8",
  "ultralytics_version": "8.4.115"
}