[ { "metadataOutputVersion" : "3.0", "storagePrecision" : "Float16", "outputSchema" : [ { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 1 × 16192)", "shortDescription" : "", "shape" : "[1, 1, 16192]", "name" : "var_2129", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 1 × 768)", "shortDescription" : "", "shape" : "[1, 1, 768]", "name" : "input", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k_1", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v_1", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_169", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k_3", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v_3", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_339", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k_5", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v_5", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_509", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k_7", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v_7", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_679", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k_9", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v_9", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_849", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k_11", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v_11", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_1019", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k_13", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v_13", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_1189", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k_15", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v_15", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_1359", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k_17", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v_17", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_1529", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k_19", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v_19", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_1699", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k_21", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v_21", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_1869", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_k", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "new_v", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "var_2039", "type" : "MultiArray" } ], "modelParameters" : [ ], "specificationVersion" : 8, "mlProgramOperationTypeHistogram" : { "Ios17.equal" : 12, "Ios17.reshape" : 84, "Ios16.softmax" : 24, "Ios17.matmul" : 48, "Ios17.transpose" : 120, "Select" : 1, "Ios17.expandDims" : 3, "Ios17.add" : 98, "Tile" : 24, "Ios17.sliceByIndex" : 60, "Ios17.lessEqual" : 12, "Ios17.gather" : 1, "Ios17.layerNorm" : 49, "Ios16.gelu" : 12, "Ios17.sub" : 25, "Ios17.conv" : 24, "Ios17.cast" : 29, "Ios17.linear" : 61, "Ios17.greaterEqual" : 1, "Ios17.mul" : 85 }, "computePrecision" : "Mixed (Float16, Int16, Int32)", "isUpdatable" : "0", "stateSchema" : [ ], "availability" : { "macOS" : "14.0", "tvOS" : "17.0", "visionOS" : "1.0", "watchOS" : "10.0", "iOS" : "17.0", "macCatalyst" : "17.0" }, "modelType" : { "name" : "MLModelType_mlProgram" }, "userDefinedMetadata" : { "com.github.apple.coremltools.conversion_date" : "2026-04-25", "com.github.apple.coremltools.source" : "torch==2.10.0", "com.github.apple.coremltools.version" : "9.0", "com.github.apple.coremltools.source_dialect" : "TorchScript" }, "inputSchema" : [ { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 1 × 768)", "shortDescription" : "", "shape" : "[1, 1, 768]", "name" : "audio_embed", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 256 × 768)", "shortDescription" : "", "shape" : "[1, 256, 768]", "name" : "encoder_output", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float32", "formattedType" : "MultiArray (Float32 1 × 256)", "shortDescription" : "", "shape" : "[1, 256]", "name" : "encoder_mask", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k0", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v0", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position0", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k1", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v1", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position1", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k2", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v2", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position2", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k3", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v3", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position3", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k4", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v4", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position4", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k5", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v5", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position5", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k6", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v6", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position6", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k7", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v7", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position7", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k8", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v8", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position8", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k9", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v9", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position9", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k10", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v10", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position10", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_k11", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 512 × 12 × 64)", "shortDescription" : "", "shape" : "[1, 512, 12, 64]", "name" : "cache_v11", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1)", "shortDescription" : "", "shape" : "[1]", "name" : "position11", "type" : "MultiArray" } ], "generatedClassName" : "decoder_step", "method" : "predict" } ]