[ { "metadataOutputVersion" : "3.0", "storagePrecision" : "Float16", "outputSchema" : [ { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Int32", "formattedType" : "MultiArray (Int32 1 × 1)", "shortDescription" : "", "shape" : "[1, 1]", "name" : "should_continue", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Int32", "formattedType" : "MultiArray (Int32 1 × 16)", "shortDescription" : "", "shape" : "[1, 16]", "name" : "frame_tokens", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 768)", "shortDescription" : "", "shape" : "[1, 768]", "name" : "next_global_embedding", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 768 × 1)", "shortDescription" : "", "shape" : "[1, 768, 1]", "name" : "codec_latent", "type" : "MultiArray" } ], "modelParameters" : [ ], "specificationVersion" : 9, "functions" : [ { "inputSchema" : [ { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 768)", "shortDescription" : "", "shape" : "[1, 768]", "name" : "global_hidden", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Int32", "formattedType" : "MultiArray (Int32 1 × 16 × 1024)", "shortDescription" : "", "shape" : "[1, 16, 1024]", "name" : "repetition_seen", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float32", "formattedType" : "MultiArray (Float32 1)", "shortDescription" : "", "shape" : "[1]", "name" : "assistant_random", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float32", "formattedType" : "MultiArray (Float32 1 × 16)", "shortDescription" : "", "shape" : "[1, 16]", "name" : "audio_random", "type" : "MultiArray" } ], "computePrecision" : "Mixed (Float16, Int32, UInt16)", "storagePrecision" : "Float16", "stateSchema" : [ ], "outputSchema" : [ { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Int32", "formattedType" : "MultiArray (Int32 1 × 1)", "shortDescription" : "", "shape" : "[1, 1]", "name" : "should_continue", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Int32", "formattedType" : "MultiArray (Int32 1 × 16)", "shortDescription" : "", "shape" : "[1, 16]", "name" : "frame_tokens", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 768)", "shortDescription" : "", "shape" : "[1, 768]", "name" : "next_global_embedding", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 768 × 1)", "shortDescription" : "", "shape" : "[1, 768, 1]", "name" : "codec_latent", "type" : "MultiArray" } ], "name" : "sample", "mlProgramOperationTypeHistogram" : { "Ios18.softmax" : 50, "Ios18.mul" : 165, "Ios18.conv" : 17, "Ios18.matmul" : 34, "Split" : 17, "Ios18.gatherAlongAxis" : 16, "Select" : 49, "Tile" : 1, "Ios18.add" : 99, "Ios16.reduceSum" : 16, "Ios18.layerNorm" : 51, "Ios16.fillLike" : 16, "Ios18.reshape" : 185, "Ios18.gather" : 32, "Ios18.linear" : 85, "Ios18.gelu" : 17, "Ios18.concat" : 32, "Ios18.topk" : 16, "Ios18.transpose" : 68, "Stack" : 35, "Ios18.cast" : 115, "Ios18.less" : 48, "Ios18.sub" : 16, "Ios18.lessEqual" : 1, "Ios18.sliceByIndex" : 118, "Ios16.cumsum" : 32, "Ios18.clip" : 16 } }, { "inputSchema" : [ { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Int32", "formattedType" : "MultiArray (Int32 1 × 128 × 17)", "shortDescription" : "", "shape" : "[1, 128, 17]", "name" : "input_ids", "type" : "MultiArray" } ], "computePrecision" : "Mixed (Float16, Int16, Int32, UInt16)", "storagePrecision" : "Mixed (Float16, Int32)", "stateSchema" : [ ], "outputSchema" : [ { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 128 × 768)", "shortDescription" : "", "shape" : "[1, 128, 768]", "name" : "embeddings", "type" : "MultiArray" } ], "name" : "embed128", "mlProgramOperationTypeHistogram" : { "Select" : 17, "Ios18.expandDims" : 16, "Ios18.gather" : 17, "Ios18.mul" : 16, "Ios18.notEqual" : 16, "Ios18.cast" : 35, "Ios18.sliceByIndex" : 17, "Ios18.greaterEqual" : 1, "Ios18.add" : 17 } }, { "inputSchema" : [ { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Int32", "formattedType" : "MultiArray (Int32 1 × 256 × 17)", "shortDescription" : "", "shape" : "[1, 256, 17]", "name" : "input_ids", "type" : "MultiArray" } ], "computePrecision" : "Mixed (Float16, Int16, Int32, UInt16)", "storagePrecision" : "Mixed (Float16, Int32)", "stateSchema" : [ ], "outputSchema" : [ { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 256 × 768)", "shortDescription" : "", "shape" : "[1, 256, 768]", "name" : "embeddings", "type" : "MultiArray" } ], "name" : "embed256", "mlProgramOperationTypeHistogram" : { "Select" : 17, "Ios18.expandDims" : 16, "Ios18.gather" : 17, "Ios18.mul" : 16, "Ios18.notEqual" : 16, "Ios18.cast" : 35, "Ios18.sliceByIndex" : 17, "Ios18.greaterEqual" : 1, "Ios18.add" : 17 } } ], "mlProgramOperationTypeHistogram" : { "Ios18.softmax" : 50, "Ios18.mul" : 165, "Ios18.conv" : 17, "Ios18.matmul" : 34, "Split" : 17, "Ios18.gatherAlongAxis" : 16, "Select" : 49, "Tile" : 1, "Ios18.add" : 99, "Ios16.reduceSum" : 16, "Ios18.layerNorm" : 51, "Ios16.fillLike" : 16, "Ios18.reshape" : 185, "Ios18.gather" : 32, "Ios18.linear" : 85, "Ios18.gelu" : 17, "Ios18.concat" : 32, "Ios18.topk" : 16, "Ios18.transpose" : 68, "Stack" : 35, "Ios18.cast" : 115, "Ios18.less" : 48, "Ios18.sub" : 16, "Ios18.lessEqual" : 1, "Ios18.sliceByIndex" : 118, "Ios16.cumsum" : 32, "Ios18.clip" : 16 }, "isUpdatable" : "0", "stateSchema" : [ ], "availability" : { "macOS" : "15.0", "tvOS" : "18.0", "visionOS" : "2.0", "watchOS" : "11.0", "iOS" : "18.0", "macCatalyst" : "18.0" }, "computePrecision" : "Mixed (Float16, Int32, UInt16)", "modelType" : { "name" : "MLModelType_mlProgram" }, "inputSchema" : [ { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float16", "formattedType" : "MultiArray (Float16 1 × 768)", "shortDescription" : "", "shape" : "[1, 768]", "name" : "global_hidden", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Int32", "formattedType" : "MultiArray (Int32 1 × 16 × 1024)", "shortDescription" : "", "shape" : "[1, 16, 1024]", "name" : "repetition_seen", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float32", "formattedType" : "MultiArray (Float32 1)", "shortDescription" : "", "shape" : "[1]", "name" : "assistant_random", "type" : "MultiArray" }, { "hasShapeFlexibility" : "0", "isOptional" : "0", "dataType" : "Float32", "formattedType" : "MultiArray (Float32 1 × 16)", "shortDescription" : "", "shape" : "[1, 16]", "name" : "audio_random", "type" : "MultiArray" } ], "defaultFunctionName" : "sample", "generatedClassName" : "MossLocal", "userDefinedMetadata" : { }, "method" : "predict" } ]