{
  "runtime": {
    "model": "convaiinnovations/laya",
    "revision": "1c5edc17a7acd8701df6fc341c0d179f1c62c982",
    "identity": "convaiinnovations/laya@1c5edc17a7acd8701df6fc341c0d179f1c62c982",
    "device": "cuda",
    "inferenceDtype": "torch.bfloat16",
    "threads": 4,
    "maxTokens": 512,
    "loadMs": 36992,
    "python": "3.12.3",
    "torch": "2.14.0+cu130",
    "transformers": "5.17.0",
    "laya": "0.3.5",
    "platform": "Windows-11-10.0.26200-SP0",
    "gpu": {
      "name": "NVIDIA GeForce RTX 5070 Ti",
      "totalBytes": 17094344704,
      "capability": [
        12,
        0
      ],
      "cuda": "13.0"
    },
    "gpuMemory": {
      "allocatedBytes": 1718991360,
      "reservedBytes": 2688548864,
      "peakAllocatedBytes": 2471085568,
      "peakReservedBytes": 2688548864
    }
  },
  "cpu": "AMD Ryzen 9 3900X 12-Core Processor",
  "systemRamBytes": 34280935424,
  "rssBytes": 1924964352,
  "peakRssBytes": 3367432192,
  "note": "Working set after text-only evaluations; peak includes loading and inference. CUDA counters are this process's PyTorch allocator only, not total GPU/driver memory. Not a hardware minimum."
}