[
  {
    "metadataOutputVersion" : "3.0",
    "storagePrecision" : "Mixed (Float32, Int4)",
    "outputSchema" : [
      {
        "hasShapeFlexibility" : "0",
        "isOptional" : "0",
        "dataType" : "Float16",
        "formattedType" : "MultiArray (Float16)",
        "shortDescription" : "",
        "shape" : "[]",
        "name" : "logits",
        "type" : "MultiArray"
      }
    ],
    "modelParameters" : [

    ],
    "specificationVersion" : 9,
    "mlProgramOperationTypeHistogram" : {
      "Ios18.expandDims" : 78,
      "Range1d" : 1,
      "Ios18.cos" : 1,
      "Ios18.mul" : 398,
      "Ios18.matmul" : 1,
      "Ios18.rsqrt" : 73,
      "Ios16.reduceMean" : 73,
      "Ios18.realDiv" : 72,
      "Ios18.scaledDotProductAttention" : 36,
      "Ios18.sin" : 1,
      "Ios18.readState" : 74,
      "Tile" : 72,
      "Ios18.gather" : 113,
      "Ios18.add" : 219,
      "Ios18.writeState" : 72,
      "Ios18.sliceUpdate" : 72,
      "Ios18.reshape" : 216,
      "Shape" : 184,
      "Ios18.linear" : 253,
      "Ios18.concat" : 219,
      "Ios18.transpose" : 145,
      "Ios18.sub" : 1,
      "Ios18.pow" : 73,
      "Ios18.cast" : 149,
      "Ios18.silu" : 36,
      "Ios18.sliceByIndex" : 324,
      "Ios18.constexprBlockwiseShiftScale" : 253
    },
    "computePrecision" : "Mixed (Float16, Float32, Int32)",
    "isUpdatable" : "0",
    "stateSchema" : [
      {
        "dataType" : "Float16",
        "isOptional" : "0",
        "formattedType" : "State (Float16 36 × 1 × 2 × 2048 × 128)",
        "shortDescription" : "",
        "shape" : "[36, 1, 2, 2048, 128]",
        "name" : "key_cache",
        "type" : "State"
      },
      {
        "dataType" : "Float16",
        "isOptional" : "0",
        "formattedType" : "State (Float16 36 × 1 × 2 × 2048 × 128)",
        "shortDescription" : "",
        "shape" : "[36, 1, 2, 2048, 128]",
        "name" : "value_cache",
        "type" : "State"
      }
    ],
    "availability" : {
      "macOS" : "15.0",
      "tvOS" : "18.0",
      "visionOS" : "2.0",
      "watchOS" : "11.0",
      "iOS" : "18.0",
      "macCatalyst" : "18.0"
    },
    "modelType" : {
      "name" : "MLModelType_mlProgram"
    },
    "userDefinedMetadata" : {
      "com.github.apple.coremltools.source_dialect" : "TorchScript",
      "com.github.apple.coremltools.source" : "torch==2.5.1",
      "com.github.apple.coremltools.version" : "8.1"
    },
    "inputSchema" : [
      {
        "dataType" : "Int32",
        "hasShapeFlexibility" : "1",
        "isOptional" : "0",
        "shapeFlexibility" : "1 × 1...2048",
        "shapeRange" : "[[1, 1], [1, 2048]]",
        "formattedType" : "MultiArray (Int32 1 × 1)",
        "type" : "MultiArray",
        "shape" : "[1, 1]",
        "name" : "input_ids",
        "shortDescription" : ""
      },
      {
        "dataType" : "Float16",
        "hasShapeFlexibility" : "1",
        "isOptional" : "0",
        "shapeFlexibility" : "1 × 1 × 1...2048 × 1...2048",
        "shapeRange" : "[[1, 1], [1, 1], [1, 2048], [1, 2048]]",
        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 1)",
        "type" : "MultiArray",
        "shape" : "[1, 1, 1, 1]",
        "name" : "causal_mask",
        "shortDescription" : ""
      }
    ],
    "generatedClassName" : "Qwen2_5_3B_Instruct_4bit",
    "method" : "predict"
  }
]