benchmark/benchmark.json

{
  "parameters": 1452562,
  "counted_parameters": 1452562,
  "macs": 1961891200,
  "gmacs": 1.9618912,
  "flops_estimate": 3923782400,
  "gflops_estimate": 3.9237824,
  "mean_latency_ms": 9.863123440172785,
  "median_latency_ms": 9.747029500431381,
  "p95_latency_ms": 11.026255997421686,
  "fps": 101.38776079055964,
  "model": "outputs/finetune_direct/p30_adamw_exact/weights/best.pt",
  "model_size_mb": 3.0141544342041016,
  "batch_size": 1,
  "imgsz": 640,
  "device": "cuda:1",
  "gpu_name": "Tesla T4",
  "gpu_total_memory_mb": 14911.6875,
  "peak_gpu_memory_mb": 29.1787109375,
  "python_version": "3.12.12",
  "torch_version": "2.10.0+cu128",
  "cuda_version": "12.8",
  "ultralytics_version": "8.4.115"
}