| [ |
| { |
| "metadataOutputVersion" : "3.0", |
| "userDefinedMetadata" : { |
| "com.anemll.lut_bits" : "6", |
| "com.github.apple.coremltools.source_dialect" : "TorchScript", |
| "com.anemll.context_length" : "1024", |
| "com.github.apple.coremltools.source" : "torch==2.5.0", |
| "com.github.apple.coremltools.version" : "8.2", |
| "com.anemll.num_chunks" : "8", |
| "com.anemll.batch_size" : "256", |
| "com.anemll.info" : "Converted with Anemll v0.2.0", |
| "com.anemll.chunk_no" : "3" |
| }, |
| "availability" : { |
| "macOS" : "15.0", |
| "tvOS" : "18.0", |
| "visionOS" : "2.0", |
| "watchOS" : "11.0", |
| "iOS" : "18.0", |
| "macCatalyst" : "18.0" |
| }, |
| "inputSchema" : [ |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Float16", |
| "formattedType" : "MultiArray (Float16 1 × 1 × 4096)", |
| "shortDescription" : "", |
| "shape" : "[1, 1, 4096]", |
| "name" : "hidden_states", |
| "type" : "MultiArray" |
| }, |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Int32", |
| "formattedType" : "MultiArray (Int32 1)", |
| "shortDescription" : "", |
| "shape" : "[1]", |
| "name" : "position_ids", |
| "type" : "MultiArray" |
| }, |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Float16", |
| "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 1024)", |
| "shortDescription" : "", |
| "shape" : "[1, 1, 1, 1024]", |
| "name" : "causal_mask", |
| "type" : "MultiArray" |
| }, |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Int32", |
| "formattedType" : "MultiArray (Int32 1)", |
| "shortDescription" : "", |
| "shape" : "[1]", |
| "name" : "current_pos", |
| "type" : "MultiArray" |
| } |
| ], |
| "outputSchema" : [ |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Float16", |
| "formattedType" : "MultiArray (Float16 1 × 1 × 4096)", |
| "shortDescription" : "", |
| "shape" : "[1, 1, 4096]", |
| "name" : "output_hidden_states", |
| "type" : "MultiArray" |
| } |
| ], |
| "modelParameters" : [ |
|
|
| ], |
| "storagePrecision" : "Mixed (Float16, Palettized (13 bits), Palettized (15 bits), Palettized (17 bits))", |
| "method" : "predict", |
| "functions" : [ |
| { |
| "inputSchema" : [ |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Float16", |
| "formattedType" : "MultiArray (Float16 1 × 1 × 4096)", |
| "shortDescription" : "", |
| "shape" : "[1, 1, 4096]", |
| "name" : "hidden_states", |
| "type" : "MultiArray" |
| }, |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Int32", |
| "formattedType" : "MultiArray (Int32 1)", |
| "shortDescription" : "", |
| "shape" : "[1]", |
| "name" : "position_ids", |
| "type" : "MultiArray" |
| }, |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Float16", |
| "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 1024)", |
| "shortDescription" : "", |
| "shape" : "[1, 1, 1, 1024]", |
| "name" : "causal_mask", |
| "type" : "MultiArray" |
| }, |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Int32", |
| "formattedType" : "MultiArray (Int32 1)", |
| "shortDescription" : "", |
| "shape" : "[1]", |
| "name" : "current_pos", |
| "type" : "MultiArray" |
| } |
| ], |
| "computePrecision" : "Mixed (Float16, Int32)", |
| "storagePrecision" : "Mixed (Float16, Palettized (13 bits), Palettized (15 bits), Palettized (17 bits))", |
| "stateSchema" : [ |
| { |
| "dataType" : "Float16", |
| "isOptional" : "0", |
| "formattedType" : "State (Float16 64 × 8 × 1024 × 128)", |
| "shortDescription" : "", |
| "shape" : "[64, 8, 1024, 128]", |
| "name" : "model_model_kv_cache_0", |
| "type" : "State" |
| } |
| ], |
| "outputSchema" : [ |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Float16", |
| "formattedType" : "MultiArray (Float16 1 × 1 × 4096)", |
| "shortDescription" : "", |
| "shape" : "[1, 1, 4096]", |
| "name" : "output_hidden_states", |
| "type" : "MultiArray" |
| } |
| ], |
| "name" : "infer", |
| "mlProgramOperationTypeHistogram" : { |
| "Ios18.expandDims" : 16, |
| "Ios18.mul" : 40, |
| "Ios18.matmul" : 8, |
| "Identity" : 1, |
| "Ios16.reduceMean" : 8, |
| "Ios18.exp" : 4, |
| "Ios18.realDiv" : 4, |
| "Ios18.greaterEqual" : 1, |
| "Select" : 1, |
| "Ios18.readState" : 9, |
| "Tile" : 8, |
| "Ios18.gather" : 2, |
| "Ios18.add" : 22, |
| "Ios18.layerNorm" : 8, |
| "Ios18.sliceUpdate" : 8, |
| "Ios18.writeState" : 8, |
| "Ios18.reshape" : 26, |
| "Ios16.reduceMax" : 4, |
| "Ios16.reduceSum" : 4, |
| "Ios18.constexprLutToDense" : 28, |
| "Ios18.conv" : 24, |
| "Ios18.concat" : 24, |
| "Ios18.transpose" : 16, |
| "Ios18.sub" : 20, |
| "Ios18.linear" : 4, |
| "Ios18.silu" : 4, |
| "Ios18.sliceByIndex" : 26, |
| "Ios18.squeeze" : 12 |
| } |
| }, |
| { |
| "inputSchema" : [ |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Float16", |
| "formattedType" : "MultiArray (Float16 1 × 256 × 4096)", |
| "shortDescription" : "", |
| "shape" : "[1, 256, 4096]", |
| "name" : "hidden_states", |
| "type" : "MultiArray" |
| }, |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Int32", |
| "formattedType" : "MultiArray (Int32 256)", |
| "shortDescription" : "", |
| "shape" : "[256]", |
| "name" : "position_ids", |
| "type" : "MultiArray" |
| }, |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Float16", |
| "formattedType" : "MultiArray (Float16 1 × 1 × 256 × 1024)", |
| "shortDescription" : "", |
| "shape" : "[1, 1, 256, 1024]", |
| "name" : "causal_mask", |
| "type" : "MultiArray" |
| }, |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Int32", |
| "formattedType" : "MultiArray (Int32 1)", |
| "shortDescription" : "", |
| "shape" : "[1]", |
| "name" : "current_pos", |
| "type" : "MultiArray" |
| } |
| ], |
| "computePrecision" : "Mixed (Float16, Int32)", |
| "storagePrecision" : "Mixed (Float16, Palettized (13 bits), Palettized (15 bits), Palettized (17 bits))", |
| "stateSchema" : [ |
| { |
| "dataType" : "Float16", |
| "isOptional" : "0", |
| "formattedType" : "State (Float16 64 × 8 × 1024 × 128)", |
| "shortDescription" : "", |
| "shape" : "[64, 8, 1024, 128]", |
| "name" : "model_model_kv_cache_0", |
| "type" : "State" |
| } |
| ], |
| "outputSchema" : [ |
| { |
| "hasShapeFlexibility" : "0", |
| "isOptional" : "0", |
| "dataType" : "Float16", |
| "formattedType" : "MultiArray (Float16 1 × 256 × 4096)", |
| "shortDescription" : "", |
| "shape" : "[1, 256, 4096]", |
| "name" : "output_hidden_states", |
| "type" : "MultiArray" |
| } |
| ], |
| "name" : "prefill", |
| "mlProgramOperationTypeHistogram" : { |
| "Ios18.expandDims" : 16, |
| "Ios18.mul" : 40, |
| "Ios18.matmul" : 8, |
| "Ios16.reduceMean" : 8, |
| "Ios18.exp" : 4, |
| "Ios18.realDiv" : 4, |
| "Ios18.greaterEqual" : 1, |
| "Select" : 1, |
| "Ios18.readState" : 9, |
| "Tile" : 8, |
| "Ios18.gather" : 2, |
| "Ios18.add" : 22, |
| "Ios18.layerNorm" : 8, |
| "Ios18.sliceUpdate" : 8, |
| "Ios18.writeState" : 8, |
| "Ios18.reshape" : 34, |
| "Ios16.reduceMax" : 4, |
| "Ios16.reduceSum" : 4, |
| "Ios18.constexprLutToDense" : 28, |
| "Ios18.conv" : 24, |
| "Ios18.concat" : 24, |
| "Ios18.transpose" : 30, |
| "Ios18.sub" : 20, |
| "Ios18.linear" : 4, |
| "Ios18.silu" : 4, |
| "Ios18.sliceByIndex" : 26, |
| "Ios18.squeeze" : 12 |
| } |
| } |
| ], |
| "version" : "0.2.0", |
| "isUpdatable" : "0", |
| "defaultFunctionName" : "infer", |
| "specificationVersion" : 9, |
| "stateSchema" : [ |
| { |
| "dataType" : "Float16", |
| "isOptional" : "0", |
| "formattedType" : "State (Float16 64 × 8 × 1024 × 128)", |
| "shortDescription" : "", |
| "shape" : "[64, 8, 1024, 128]", |
| "name" : "model_model_kv_cache_0", |
| "type" : "State" |
| } |
| ], |
| "computePrecision" : "Mixed (Float16, Int32)", |
| "mlProgramOperationTypeHistogram" : { |
| "Ios18.expandDims" : 16, |
| "Ios18.mul" : 40, |
| "Ios18.matmul" : 8, |
| "Identity" : 1, |
| "Ios16.reduceMean" : 8, |
| "Ios18.exp" : 4, |
| "Ios18.realDiv" : 4, |
| "Ios18.greaterEqual" : 1, |
| "Select" : 1, |
| "Ios18.readState" : 9, |
| "Tile" : 8, |
| "Ios18.gather" : 2, |
| "Ios18.add" : 22, |
| "Ios18.layerNorm" : 8, |
| "Ios18.sliceUpdate" : 8, |
| "Ios18.writeState" : 8, |
| "Ios18.reshape" : 26, |
| "Ios16.reduceMax" : 4, |
| "Ios16.reduceSum" : 4, |
| "Ios18.constexprLutToDense" : 28, |
| "Ios18.conv" : 24, |
| "Ios18.concat" : 24, |
| "Ios18.transpose" : 16, |
| "Ios18.sub" : 20, |
| "Ios18.linear" : 4, |
| "Ios18.silu" : 4, |
| "Ios18.sliceByIndex" : 26, |
| "Ios18.squeeze" : 12 |
| }, |
| "shortDescription" : "Anemll Model: Multifunction FFN+Prefill", |
| "generatedClassName" : "DeepSeek_FFN_PF_lut6_chunk_03of08", |
| "author" : "Converted with Anemll v0.2.0", |
| "modelType" : { |
| "name" : "MLModelType_mlProgram" |
| } |
| } |
| ] |
