| [ |
| { |
| "metadataOutputVersion" : "3.0", |
| "storagePrecision" : "Mixed (Float16, Int4)", |
| "outputSchema" : [ |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Float16", |
| "formattedType" : "MultiArray (Float16)", |
| "shortDescription" : "", |
| "shape" : "[]", |
| "name" : "logits", |
| "type" : "MultiArray" |
| } |
| ], |
| "modelParameters" : [ |
|
|
| ], |
| "specificationVersion" : 9, |
| "mlProgramOperationTypeHistogram" : { |
| "Ios18.expandDims" : 62, |
| "Range1d" : 1, |
| "Ios18.cos" : 1, |
| "Ios18.mul" : 310, |
| "Ios18.matmul" : 1, |
| "Ios18.rsqrt" : 57, |
| "Ios16.reduceMean" : 57, |
| "Ios18.realDiv" : 56, |
| "Ios18.scaledDotProductAttention" : 28, |
| "Ios18.sin" : 1, |
| "Ios18.readState" : 58, |
| "Tile" : 56, |
| "Ios18.gather" : 89, |
| "Ios18.add" : 171, |
| "Ios18.writeState" : 56, |
| "Ios18.sliceUpdate" : 56, |
| "Ios18.reshape" : 168, |
| "Shape" : 144, |
| "Ios18.linear" : 197, |
| "Ios18.concat" : 171, |
| "Ios18.transpose" : 113, |
| "Ios18.sub" : 1, |
| "Ios18.pow" : 57, |
| "Ios18.cast" : 177, |
| "Ios18.silu" : 28, |
| "Ios18.sliceByIndex" : 252, |
| "Ios18.constexprBlockwiseShiftScale" : 197 |
| }, |
| "computePrecision" : "Mixed (Float16, Float32, Int16, Int32, UInt16)", |
| "isUpdatable" : "0", |
| "stateSchema" : [ |
| { |
| "dataType" : "Float16", |
| "isOptional" : "0", |
| "formattedType" : "State (Float16 28 × 1 × 2 × 2048 × 128)", |
| "shortDescription" : "", |
| "shape" : "[28, 1, 2, 2048, 128]", |
| "name" : "key_cache", |
| "type" : "State" |
| }, |
| { |
| "dataType" : "Float16", |
| "isOptional" : "0", |
| "formattedType" : "State (Float16 28 × 1 × 2 × 2048 × 128)", |
| "shortDescription" : "", |
| "shape" : "[28, 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_Coder_1_5B_Instruct_4bit", |
| "method" : "predict" |
| } |
| ] |
