{
  "schemaVersion": 1,
  "createdUtc": "2026-08-26T05:13:12.099411+00:00",
  "experiment": "UD-IQ4_XS-Unsloth",
  "status": "complete",
  "models": {
    "native": {
      "path": "<REPO_ROOT>/Qwen3.8-27B-Quark-AWQ-MXFP4-native.gguf",
      "bytes": 18892854880,
      "decimalGB": 18.89285488,
      "GiB": 17.595342248678207,
      "sha256": "1a37c48570811215fa1a9d2a493211293ca375d17287771e85d28642946e88fc"
    },
    "candidate": {
      "path": "<REPO_ROOT>/Qwen3.8-27B-Quark-MXFP4-UD-IQ4_XS-Unsloth.gguf",
      "bytes": 13887481760,
      "decimalGB": 13.88748176,
      "GiB": 12.933725267648697,
      "sha256": "892cd2ac8aebc6f4d7cdf4d5fb0b417507a8ba9df7f038d45b9185835122cb03"
    },
    "savings": {
      "bytes": 5005373120,
      "decimalGB": 5.00537312,
      "GiB": 4.6616169810295105,
      "percentOfNative": 26.49347148322562
    }
  },
  "tensorAudit": {
    "verificationPassed": true,
    "tensorCount": 866,
    "changedTensorCount": 91,
    "unchangedPayloadsVerifiedByteExact": 775,
    "nativeMxfp4PreservedCount": 415,
    "nativeMxfp4TensorCount": 496,
    "effectiveTypeCounts": {
      "F32": 360,
      "IQ2_S": 7,
      "IQ2_XS": 2,
      "IQ3_S": 46,
      "IQ3_XXS": 17,
      "MXFP4": 415,
      "Q2_K": 1,
      "Q3_K": 9,
      "Q5_K": 1,
      "Q6_K": 6,
      "Q8_0": 2
    }
  },
  "benchmark": {
    "domain": "general",
    "tokenTarget": 32768,
    "chunks": 15,
    "context": 2048,
    "threads": 4,
    "gpuLayers": 0,
    "gpuVisible": false,
    "nativeStandalonePpl": 5.8033,
    "nativeStandalonePplError": 0.11146,
    "paired": {
      "candidatePpl": 5.82498,
      "basePpl": 5.801511,
      "pplDifference": 0.023469,
      "pplRatio": 1.004045,
      "logPplRatio": 0.004037,
      "correlationPercent": 99.78
    },
    "kld": {
      "mean": 0.009018,
      "standardError": 0.000191,
      "maximum": 0.904095,
      "p99_9": 0.31015,
      "p99": 0.085121,
      "p95": 0.030207,
      "p90": 0.019024,
      "median": 0.003997
    },
    "tokenProbability": {
      "meanDeltaPercent": -0.077,
      "rmsDeltaPercent": 2.827,
      "sameTopPercent": 95.758
    },
    "scope": {
      "languageLogits": true,
      "visionProjectorEvaluated": false,
      "mtpTensorsEvaluated": false,
      "note": "llama-perplexity logged block 64 MTP tensors as unused; ordinary KLD primarily measures the changed token embedding and output tensors"
    }
  },
  "imatrixApplicability": {
    "supplied": true,
    "label": "Unsloth",
    "sha256": "0ee5b10bd0c2fa2127c6f4b43dbfe1efd71e383b63217af9dade1de36599f1c1",
    "entries": 496,
    "changedTensorsWithEntries": 81,
    "changedTensorCount": 91,
    "note": "Coverage is inferred from llama.cpp's per-conversion missing-weight messages; tensors present in the imatrix are supplied to the selected quantizer."
  },
  "artifacts": {
    "corpus": {
      "path": "<REPO_ROOT>/Experiments/MQ-IQ4_XS_1-Generic/benchmark/_ppl_corpora/ppl_corpus_general.txt",
      "bytes": 131514,
      "sha256": "5d38d98dce15f54e9a1a926187b6058e65cd8b3dd9b5cc2729b0a5fd249228b0"
    },
    "referenceLogits": {
      "path": "<REPO_ROOT>/Experiments/MQ-IQ4_XS_1-Generic/benchmark/native-reference/logits/kld_logits_general.bin",
      "bytes": 7621186460,
      "sha256": "604db2df7253cba18a6c0a569ed819ee3065d4ac20cd94ae72f4329e7bba8516"
    },
    "nativeLog": {
      "path": "<REPO_ROOT>/Experiments/MQ-IQ4_XS_1-Generic/benchmark/native-reference/perplexity_general.log",
      "sha256": "bade3445565a354969f7515045507b854bee83f6695eec7490b70eab9c6eebc2"
    },
    "candidateLog": {
      "path": "<REPO_ROOT>/Experiments/UD-IQ4_XS-Unsloth/benchmark/candidate/perplexity_general.log",
      "sha256": "400ed964d45a1a97f307ef157caf0c2e67bdd9f975c71c65fbc0332315a02d3d"
    },
    "quantizationLog": {
      "path": "<REPO_ROOT>/Experiments/UD-IQ4_XS-Unsloth/quantization.log",
      "sha256": "18630ebadff2fe8b4a10c39771bbc9705914fc7cd83b8965209b1b007c36b0bd"
    }
  }
}