{
  "protocolVersion": "osai-ctxfit/1",
  "createdAt": "2026-10-20T06:29:20.353Z",
  "tool": "scripts/lab/context-fit-sweep.mjs",
  "runtime": {
    "name": "ollama",
    "version": "0.35.1",
    "endpoint": "loopback"
  },
  "system": {
    "cpu": {
      "model": "Intel(R) Core(TM) i7-14700K",
      "logicalCores": 28
    },
    "memory": {
      "totalBytes": 34018738176
    },
    "os": {
      "platform": "win32",
      "release": "10.0.26200"
    },
    "gpu": {
      "model": "NVIDIA GeForce RTX 4070 Ti",
      "totalVramBytes": 12878610432,
      "driverVersion": "617.42",
      "provider": "nvidia-smi"
    }
  },
  "model": {
    "identifier": "phi4:latest",
    "digest": "ac896e5b8b34a1f4efa7b14d7520725140d5512484457fab45d2a4ea14c69dba",
    "weightsBytes": 9053116391,
    "family": "phi3",
    "parameterSize": "14.7B",
    "quantization": "Q4_K_M",
    "trainedContextTokens": 16384
  },
  "settings": {
    "contexts": [
      4096,
      8192,
      16384
    ],
    "skippedContexts": [
      32768
    ],
    "passes": 4,
    "warmups": 1,
    "numPredict": 64,
    "temperature": 0,
    "seed": 42,
    "prompt": "fixed short prompt, see PROMPT in scripts/lab/context-fit-sweep.mjs",
    "settleSeconds": 8
  },
  "steps": [
    {
      "numCtx": 4096,
      "gpuBefore": {
        "usedBytes": 1522532352,
        "freeBytes": 11356078080,
        "utilizationPercent": 25
      },
      "gpuAfterLoad": {
        "usedBytes": 11619270656,
        "freeBytes": 1259339776
      },
      "loadDurationNs": 6055193900,
      "ps": {
        "sizeBytes": 9734685981,
        "sizeVramBytes": 9734685981,
        "hostBytes": 0,
        "contextLength": 4096
      },
      "passes": [
        {
          "warmup": true,
          "evalCount": 64,
          "evalDurationNs": 1400308000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 99644000
        },
        {
          "warmup": false,
          "evalCount": 64,
          "evalDurationNs": 1412954000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 22388000
        },
        {
          "warmup": false,
          "evalCount": 64,
          "evalDurationNs": 1408925000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 22538000
        },
        {
          "warmup": false,
          "evalCount": 64,
          "evalDurationNs": 1423941000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 23462000
        }
      ],
      "derived": {
        "measuredPasses": 3,
        "generationTokensPerSecond": 45.295175922216856,
        "fullyInVram": true
      }
    },
    {
      "numCtx": 8192,
      "gpuBefore": {
        "usedBytes": 1695547392,
        "freeBytes": 11183063040,
        "utilizationPercent": 34
      },
      "gpuAfterLoad": {
        "usedBytes": 12103712768,
        "freeBytes": 774897664
      },
      "loadDurationNs": 6782057000,
      "ps": {
        "sizeBytes": 11018877007,
        "sizeVramBytes": 10430007213,
        "hostBytes": 588869794,
        "contextLength": 8192
      },
      "passes": [
        {
          "warmup": true,
          "evalCount": 64,
          "evalDurationNs": 5950257000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 107937000
        },
        {
          "warmup": false,
          "evalCount": 64,
          "evalDurationNs": 9516370000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 73209000
        },
        {
          "warmup": false,
          "evalCount": 64,
          "evalDurationNs": 8945303000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 39271000
        },
        {
          "warmup": false,
          "evalCount": 64,
          "evalDurationNs": 9036570000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 58749000
        }
      ],
      "derived": {
        "measuredPasses": 3,
        "generationTokensPerSecond": 7.082333230418179,
        "fullyInVram": false
      }
    },
    {
      "numCtx": 16384,
      "gpuBefore": {
        "usedBytes": 1604321280,
        "freeBytes": 11274289152,
        "utilizationPercent": 24
      },
      "gpuAfterLoad": {
        "usedBytes": 12002000896,
        "freeBytes": 876609536
      },
      "loadDurationNs": 6758777000,
      "ps": {
        "sizeBytes": 12675627087,
        "sizeVramBytes": 10258596494,
        "hostBytes": 2417030593,
        "contextLength": 16384
      },
      "passes": [
        {
          "warmup": true,
          "evalCount": 64,
          "evalDurationNs": 2568540000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 170414000
        },
        {
          "warmup": false,
          "evalCount": 64,
          "evalDurationNs": 2538328000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 45236000
        },
        {
          "warmup": false,
          "evalCount": 64,
          "evalDurationNs": 2477201000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 43067000
        },
        {
          "warmup": false,
          "evalCount": 64,
          "evalDurationNs": 2479405000,
          "promptEvalCount": 35,
          "promptEvalDurationNs": 39518000
        }
      ],
      "derived": {
        "measuredPasses": 3,
        "generationTokensPerSecond": 25.812644565934168,
        "fullyInVram": false
      }
    }
  ]
}
