| [ | |
| { | |
| "metadataOutputVersion" : "3.0", | |
| "storagePrecision" : "Mixed (Float16, Palettized (4 bits), Palettized (6 bits))", | |
| "outputSchema" : [ | |
| { | |
| "hasShapeFlexibility" : "0", | |
| "isOptional" : "0", | |
| "dataType" : "Float16", | |
| "formattedType" : "MultiArray (Float16 1 × 1 × 51866)", | |
| "shortDescription" : "", | |
| "shape" : "[1, 1, 51866]", | |
| "name" : "logits", | |
| "type" : "MultiArray" | |
| }, | |
| { | |
| "hasShapeFlexibility" : "0", | |
| "isOptional" : "0", | |
| "dataType" : "Float16", | |
| "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 1)", | |
| "shortDescription" : "", | |
| "shape" : "[1, 5120, 1, 1]", | |
| "name" : "key_cache_updates", | |
| "type" : "MultiArray" | |
| }, | |
| { | |
| "hasShapeFlexibility" : "0", | |
| "isOptional" : "0", | |
| "dataType" : "Float16", | |
| "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 1)", | |
| "shortDescription" : "", | |
| "shape" : "[1, 5120, 1, 1]", | |
| "name" : "value_cache_updates", | |
| "type" : "MultiArray" | |
| }, | |
| { | |
| "hasShapeFlexibility" : "0", | |
| "isOptional" : "0", | |
| "dataType" : "Float16", | |
| "formattedType" : "MultiArray (Float16 1 × 1500)", | |
| "shortDescription" : "", | |
| "shape" : "[1, 1500]", | |
| "name" : "alignment_heads_weights", | |
| "type" : "MultiArray" | |
| } | |
| ], | |
| "modelParameters" : [ | |
| ], | |
| "specificationVersion" : 7, | |
| "mlProgramOperationTypeHistogram" : { | |
| "Transpose" : 1, | |
| "Ios16.gather" : 2, | |
| "Squeeze" : 1, | |
| "Ios16.reduceMean" : 1, | |
| "Ios16.softmax" : 8, | |
| "Split" : 2, | |
| "Ios16.linear" : 1, | |
| "Ios16.add" : 65, | |
| "Concat" : 3, | |
| "ExpandDims" : 6, | |
| "Ios16.sub" : 1, | |
| "Ios16.conv" : 120, | |
| "Ios16.gelu" : 4, | |
| "Ios16.constexprLutToDense" : 40, | |
| "Ios16.matmul" : 16, | |
| "Ios16.layerNorm" : 13, | |
| "Ios16.batchNorm" : 13, | |
| "Ios16.reshape" : 32, | |
| "SliceByIndex" : 12, | |
| "Ios16.mul" : 24 | |
| }, | |
| "computePrecision" : "Mixed (Float16, Int32)", | |
| "isUpdatable" : "0", | |
| "availability" : { | |
| "macOS" : "13.0", | |
| "tvOS" : "16.0", | |
| "visionOS" : "1.0", | |
| "watchOS" : "9.0", | |
| "iOS" : "16.0", | |
| "macCatalyst" : "16.0" | |
| }, | |
| "modelType" : { | |
| "name" : "MLModelType_mlProgram" | |
| }, | |
| "userDefinedMetadata" : { | |
| "com.github.apple.coremltools.source_dialect" : "TorchScript", | |
| "com.github.apple.coremltools.version" : "8.0", | |
| "com.github.apple.coremltools.source" : "torch==2.4.1" | |
| }, | |
| "inputSchema" : [ | |
| { | |
| "hasShapeFlexibility" : "0", | |
| "isOptional" : "0", | |
| "dataType" : "Int32", | |
| "formattedType" : "MultiArray (Int32 1)", | |
| "shortDescription" : "", | |
| "shape" : "[1]", | |
| "name" : "input_ids", | |
| "type" : "MultiArray" | |
| }, | |
| { | |
| "hasShapeFlexibility" : "0", | |
| "isOptional" : "0", | |
| "dataType" : "Int32", | |
| "formattedType" : "MultiArray (Int32 1)", | |
| "shortDescription" : "", | |
| "shape" : "[1]", | |
| "name" : "cache_length", | |
| "type" : "MultiArray" | |
| }, | |
| { | |
| "hasShapeFlexibility" : "0", | |
| "isOptional" : "0", | |
| "dataType" : "Float16", | |
| "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 448)", | |
| "shortDescription" : "", | |
| "shape" : "[1, 5120, 1, 448]", | |
| "name" : "key_cache", | |
| "type" : "MultiArray" | |
| }, | |
| { | |
| "hasShapeFlexibility" : "0", | |
| "isOptional" : "0", | |
| "dataType" : "Float16", | |
| "formattedType" : "MultiArray (Float16 1 × 5120 × 1 × 448)", | |
| "shortDescription" : "", | |
| "shape" : "[1, 5120, 1, 448]", | |
| "name" : "value_cache", | |
| "type" : "MultiArray" | |
| }, | |
| { | |
| "hasShapeFlexibility" : "0", | |
| "isOptional" : "0", | |
| "dataType" : "Float16", | |
| "formattedType" : "MultiArray (Float16 1 × 448)", | |
| "shortDescription" : "", | |
| "shape" : "[1, 448]", | |
| "name" : "kv_cache_update_mask", | |
| "type" : "MultiArray" | |
| }, | |
| { | |
| "hasShapeFlexibility" : "0", | |
| "isOptional" : "0", | |
| "dataType" : "Float16", | |
| "formattedType" : "MultiArray (Float16 1 × 1280 × 1 × 1500)", | |
| "shortDescription" : "", | |
| "shape" : "[1, 1280, 1, 1500]", | |
| "name" : "encoder_output_embeds", | |
| "type" : "MultiArray" | |
| }, | |
| { | |
| "hasShapeFlexibility" : "0", | |
| "isOptional" : "0", | |
| "dataType" : "Float16", | |
| "formattedType" : "MultiArray (Float16 1 × 448)", | |
| "shortDescription" : "", | |
| "shape" : "[1, 448]", | |
| "name" : "decoder_key_padding_mask", | |
| "type" : "MultiArray" | |
| } | |
| ], | |
| "generatedClassName" : "TextDecoder_mixedBitPalettized_4_1_bit", | |
| "method" : "predict" | |
| } | |
| ] |