{
  "probe": "greedy_generation",
  "provider": "CUDAExecutionProvider",
  "session_load_seconds": 28.961127704940736,
  "step_seconds": [
    0.18181518698111176,
    0.030539471190422773,
    0.028461792040616274,
    0.028141363989561796,
    0.028162146918475628,
    0.028294692980125546,
    0.028191918972879648,
    0.028165633091703057
  ],
  "prompt_token_ids": [
    7985,
    825,
    2805,
    11652,
    911,
    6197,
    43304,
    13
  ],
  "generated_token_ids": [
    7985,
    825,
    2805,
    11652,
    911,
    6197,
    43304,
    13,
    6197,
    43304,
    374,
    264,
    5411,
    1787,
    30774,
    12626
  ],
  "new_token_ids": [
    6197,
    43304,
    374,
    264,
    5411,
    1787,
    30774,
    12626
  ],
  "text": "Write one short sentence about ONNX. ONNX is a popular open-source framework",
  "finite_logits": true,
  "source_model": "Qwen/Qwen2.5-0.5B-Instruct",
  "source_revision": "7ae557604adf67be50417f59c2c2f167def9a775",
  "versions": {
    "python": "3.12.9",
    "onnxruntime": "1.29.0",
    "onnx_ir": "1.0.0",
    "transformers": "5.15.1",
    "torch": "2.13.0+cu130",
    "numpy": "2.5.2",
    "providers_available": [
      "TensorrtExecutionProvider",
      "CUDAExecutionProvider",
      "CPUExecutionProvider"
    ]
  }
}