{
  "schema": "marker-gpu-trace-runtime-projection/v1",
  "phase": "incumbent_before_trial",
  "identity": {
    "model": "nvidia/GLM-5.3-Flash-FP4-Local",
    "revision": "76348ddb2be2e35e39ecfc00e75dc07703f4f3d5",
    "image": "ghcr.io/local-inference-lab/vllm:release-10152026@sha256:745feb916ee816f5409fd58386114be4f22fbede7f9c0e0a7b8c4fe8dac0fb29",
    "image_id": "sha256:745feb916ee816f5409fd58386114be4f22fbede7f9c0e0a7b8c4fe8dac0fb29",
    "state": "running",
    "launch": {
      "command_sha256": "00dda05b0fed3df771a324bc6c5e16eeacb358ce06a376f930f969c5fae7f81d",
      "entrypoint_sha256": "ae8a7ccee10b493bede97aae9d230c562a193c687fac03f5808a23074ea67a60",
      "mounts_sha256": "ef9bac739b828b5e4924fba9fad62bf97623fd0979426d89bbdde96a9741174e",
      "reasoning_parser": "glm45",
      "speculative_tokens": 3,
      "v2_model_runner": true,
      "flashinfer_sampler": true,
      "fastokens": true
    },
    "runtime_patch": null
  },
  "engine": {"running": 0, "waiting": 0, "inference_requests": 0},
  "resources": {
    "host_mem_available_bytes": 72013271040,
    "host_oom_kill_counter": 3,
    "cgroup_current_bytes": 25518227456,
    "cgroup_peak_bytes": 35379359744,
    "cgroup_oom_killed": false,
    "cgroup_events": {"low": 0, "high": 0, "max": 0, "oom": 0, "oom_kill": 0, "oom_group_kill": 0}
  },
  "projection": {
    "native_source_sha256": "f5f2a2dbe38ea383e810672b1526e4669ef9911000b0f079c5116e56841d7e9a",
    "transformations": [
      "Removed private host, container, process, path, command, environment and GPU identity data.",
      "Preserved incumbent model/image/launch identity, engine occupancy and bounded resource/OOM observations."
    ]
  }
}