[ { "metadataOutputVersion" : "3.0", "storagePrecision" : "Float16", "outputSchema" : [ { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16)", "shortDescription" : "", "shape" : "[]", "name" : "logits", "type" : "MultiArray" } ], "modelParameters" : [ ], "specificationVersion" : 9, "mlProgramOperationTypeHistogram" : { "Ios18.linear" : 49, "Ios18.readState" : 14, "Ios18.expandDims" : 7, "Ios18.sub" : 1, "Ios18.matmul" : 24, "Ios18.gelu" : 6, "Ios18.gather" : 9, "Ios18.concat" : 32, "Shape" : 8, "Ios18.add" : 31, "Ios18.sliceUpdate" : 24, "Ios18.sliceByIndex" : 49, "Ios18.layerNorm" : 19, "Ios18.cast" : 16, "Ios18.transpose" : 48, "Ios18.writeState" : 12, "Ios18.reshape" : 48, "Ios18.softmax" : 12, "Ios18.mul" : 24 }, "computePrecision" : "Mixed (Float16, Int16, Int32, UInt16)", "isUpdatable" : "0", "stateSchema" : [ { "dataType" : "Float16", "isOptional" : "0", "formattedType" : "State (Float16 6 × 1 × 448 × 512)", "shortDescription" : "", "shape" : "[6, 1, 448, 512]", "name" : "k_cache1", "type" : "State" }, { "dataType" : "Float16", "isOptional" : "0", "formattedType" : "State (Float16 6 × 1 × 448 × 512)", "shortDescription" : "", "shape" : "[6, 1, 448, 512]", "name" : "v_cache1", "type" : "State" }, { "dataType" : "Float16", "isOptional" : "0", "formattedType" : "State (Float16 6 × 1 × 1500 × 512)", "shortDescription" : "", "shape" : "[6, 1, 1500, 512]", "name" : "k_cache2", "type" : "State" }, { "dataType" : "Float16", "isOptional" : "0", "formattedType" : "State (Float16 6 × 1 × 1500 × 512)", "shortDescription" : "", "shape" : "[6, 1, 1500, 512]", "name" : "v_cache2", "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==2.4.1", "com.github.apple.coremltools.version" : "8.0" }, "inputSchema" : [ { "dataType" : "Int32", "hasShapeFlexibility" : "1", "isOptional" : "0", "shapeFlexibility" : "1 × 1...448", "shapeRange" : "[[1, 1], [1, 448]]", "formattedType" : "MultiArray (Int32 1 × 1)", "type" : "MultiArray", "shape" : "[1, 1]", "name" : "token_data", "shortDescription" : "" }, { "dataType" : "Float16", "hasShapeFlexibility" : "1", "isOptional" : "0", "shapeFlexibility" : "1 × 1...448", "shapeRange" : "[[1, 1], [1, 448]]", "formattedType" : "MultiArray (Float16 1 × 1)", "type" : "MultiArray", "shape" : "[1, 1]", "name" : "offset_mask", "shortDescription" : "" } ], "generatedClassName" : "decoder_second", "method" : "predict" } ]