{
  "name": "nvfp4",
  "quant_config": "NVFP4_MLP_ONLY_CFG",
  "base_model": "nvidia/Nemotron-3-Embed-8B-BF16",
  "base_revision": "8ca3ff382cf1de715e05acac8b553e0a084680d0",
  "n_calibration_samples": 736,
  "calibration_audit": {
    "seed": 20260803,
    "total_samples": 736,
    "fill_ratio": 0.9787,
    "per_bucket": {
      "passage/128": {
        "need": 130,
        "filled": 130
      },
      "passage/16384": {
        "need": 24,
        "filled": 24
      },
      "passage/2048": {
        "need": 120,
        "filled": 120
      },
      "passage/32": {
        "need": 30,
        "filled": 30
      },
      "passage/512": {
        "need": 160,
        "filled": 160
      },
      "passage/8192": {
        "need": 60,
        "filled": 60
      },
      "query/128": {
        "need": 70,
        "filled": 54
      },
      "query/32": {
        "need": 158,
        "filled": 158
      }
    },
    "zero_buckets": [],
    "reallocations": [
      {
        "from": "query/512",
        "n": 8,
        "to": "query/32"
      }
    ],
    "truncated": 0,
    "miracl_langs_missing": [
      "de"
    ],
    "holdout_languages": [
      "sw",
      "te"
    ],
    "status": "OK"
  },
  "module_inventory_len": 239,
  "fixture_cosines_vs_bf16": {
    "6943a1bd2fe92890": 0.991631,
    "818e9615d444ae07": 0.990915,
    "9b8079f3b1d55d58": 0.993484,
    "702fbaddfce52906": 0.991667,
    "2d7b285cfc45f158": 0.989624
  },
  "shard_sha256": {
    "model.safetensors": "17f0c0cf3c6c80477f43413168bdf90a0f6341accab9a7548f5301c457e28104"
  },
  "versions": {
    "modelopt": "0.45.0",
    "transformers": "5.14.1",
    "torch": "2.8.0+cu128"
  },
  "completion_marker": true
}