Initial commit

.gitattributes

@@ -0,0 +1,35 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,7 @@
+---
+license: apache-2.0
+---
+
+# Tape
+
+Coming soon.

medium/AudioEncoder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1a8ab7aec23b627541276af4d0bb0e4000050282e2bcb684b64584fba27a9a4
+size 243

medium/AudioEncoder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90b5e31a285d2c0af400e27a7e38838beaec50cf44da2d7534fbdec70bfcc686
+size 409

medium/AudioEncoder.mlmodelc/metadata.json

@@ -0,0 +1,73 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Palettized (8 bits), Sparse)",
+    "outputSchema" : [
+      {
+        "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"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 7,
+    "mlProgramOperationTypeHistogram" : {
+      "Concat" : 672,
+      "Ios16.mul" : 3840,
+      "Ios16.layerNorm" : 65,
+      "SliceByIndex" : 5760,
+      "Ios16.constexprLutToDense" : 194,
+      "Transpose" : 32,
+      "Ios16.einsum" : 7680,
+      "Ios16.conv" : 388,
+      "Ios16.add" : 259,
+      "Ios16.constexprSparseToDense" : 192,
+      "Ios16.softmax" : 3840,
+      "Ios16.gelu" : 34,
+      "Ios16.batchNorm" : 65
+    },
+    "computePrecision" : "Mixed (Float16, Int32)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "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.conversion_date" : "2026-03-05",
+      "com.github.apple.coremltools.source" : "torch==2.5.0",
+      "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 × 128 × 1 × 3000)",
+        "shortDescription" : "",
+        "shape" : "[1, 128, 1, 3000]",
+        "name" : "melspectrogram_features",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "AudioEncoder_mixedBitPalettized_8_bit",
+    "method" : "predict"
+  }
+]

medium/AudioEncoder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e260848b44cd73c7ce878c55504d678bd12e602cbeaa39c2596857e4941ec4b
+size 739291328

medium/MelSpectrogram.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2592a0e27cd8a703fe7c7d2a6a0fca872efebfbb8e2f0640320a89c5a553573d
+size 243

medium/MelSpectrogram.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f049e1a21b69f3617487af4198470abb6eea4a98eb8dbcaf57c9eaca640ff91
+size 390

medium/MelSpectrogram.mlmodelc/metadata.json

@@ -0,0 +1,75 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Float16",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 128 × 1 × 3000)",
+        "shortDescription" : "",
+        "shape" : "[1, 128, 1, 3000]",
+        "name" : "melspectrogram_features",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 7,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios16.reshape" : 2,
+      "Ios16.mul" : 2,
+      "SliceByIndex" : 1,
+      "Ios16.sub" : 1,
+      "Ios16.log" : 1,
+      "Ios16.square" : 2,
+      "Ios16.add" : 3,
+      "Squeeze" : 2,
+      "Ios16.matmul" : 1,
+      "Ios16.conv" : 2,
+      "Ios16.maximum" : 1,
+      "ExpandDims" : 4,
+      "Ios16.reduceMax" : 1,
+      "Identity" : 1,
+      "Pad" : 1
+    },
+    "computePrecision" : "Mixed (Float16, Int32)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "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.conversion_date" : "2026-03-04",
+      "com.github.apple.coremltools.source" : "torch==2.5.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 480000)",
+        "shortDescription" : "",
+        "shape" : "[480000]",
+        "name" : "audio",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "MelSpectrogram",
+    "method" : "predict"
+  }
+]

medium/MelSpectrogram.mlmodelc/model.mil

@@ -0,0 +1,66 @@
+program(1.0)
+[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "3500.14.1"}, {"coremlc-version", "3500.32.1"}, {"coremltools-component-torch", "2.5.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "9.0"}})]
+{
+    func main<ios16>(tensor<fp16, [480000]> audio) {
+            tensor<int32, [3]> var_10 = const()[name = tensor<string, []>("op_10"), val = tensor<int32, [3]>([1, 1, 480000])];
+            tensor<fp16, [1, 1, 480000]> input_1_cast_fp16 = reshape(shape = var_10, x = audio)[name = tensor<string, []>("input_1_cast_fp16")];
+            tensor<int32, [6]> input_3_pad_0 = const()[name = tensor<string, []>("input_3_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 200, 200])];
+            tensor<string, []> input_3_mode_0 = const()[name = tensor<string, []>("input_3_mode_0"), val = tensor<string, []>("reflect")];
+            tensor<fp16, []> const_1_to_fp16 = const()[name = tensor<string, []>("const_1_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
+            tensor<fp16, [1, 1, 480400]> input_3_cast_fp16 = pad(constant_val = const_1_to_fp16, mode = input_3_mode_0, pad = input_3_pad_0, x = input_1_cast_fp16)[name = tensor<string, []>("input_3_cast_fp16")];
+            tensor<int32, [1]> var_22 = const()[name = tensor<string, []>("op_22"), val = tensor<int32, [1]>([480400])];
+            tensor<fp16, [480400]> input_cast_fp16 = reshape(shape = var_22, x = input_3_cast_fp16)[name = tensor<string, []>("input_cast_fp16")];
+            tensor<int32, [1]> expand_dims_0_axes_0 = const()[name = tensor<string, []>("expand_dims_0_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [1, 480400]> expand_dims_0_cast_fp16 = expand_dims(axes = expand_dims_0_axes_0, x = input_cast_fp16)[name = tensor<string, []>("expand_dims_0_cast_fp16")];
+            tensor<int32, [1]> expand_dims_3 = const()[name = tensor<string, []>("expand_dims_3"), val = tensor<int32, [1]>([160])];
+            tensor<int32, [1]> expand_dims_4_axes_0 = const()[name = tensor<string, []>("expand_dims_4_axes_0"), val = tensor<int32, [1]>([1])];
+            tensor<fp16, [1, 1, 480400]> expand_dims_4_cast_fp16 = expand_dims(axes = expand_dims_4_axes_0, x = expand_dims_0_cast_fp16)[name = tensor<string, []>("expand_dims_4_cast_fp16")];
+            tensor<string, []> conv_0_pad_type_0 = const()[name = tensor<string, []>("conv_0_pad_type_0"), val = tensor<string, []>("valid")];
+            tensor<int32, [2]> conv_0_pad_0 = const()[name = tensor<string, []>("conv_0_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> conv_0_dilations_0 = const()[name = tensor<string, []>("conv_0_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, []> conv_0_groups_0 = const()[name = tensor<string, []>("conv_0_groups_0"), val = tensor<int32, []>(1)];
+            tensor<fp16, [201, 1, 400]> expand_dims_1_to_fp16 = const()[name = tensor<string, []>("expand_dims_1_to_fp16"), val = tensor<fp16, [201, 1, 400]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
+            tensor<fp16, [1, 201, 3001]> conv_0_cast_fp16 = conv(dilations = conv_0_dilations_0, groups = conv_0_groups_0, pad = conv_0_pad_0, pad_type = conv_0_pad_type_0, strides = expand_dims_3, weight = expand_dims_1_to_fp16, x = expand_dims_4_cast_fp16)[name = tensor<string, []>("conv_0_cast_fp16")];
+            tensor<string, []> conv_1_pad_type_0 = const()[name = tensor<string, []>("conv_1_pad_type_0"), val = tensor<string, []>("valid")];
+            tensor<int32, [2]> conv_1_pad_0 = const()[name = tensor<string, []>("conv_1_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> conv_1_dilations_0 = const()[name = tensor<string, []>("conv_1_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, []> conv_1_groups_0 = const()[name = tensor<string, []>("conv_1_groups_0"), val = tensor<int32, []>(1)];
+            tensor<fp16, [201, 1, 400]> expand_dims_2_to_fp16 = const()[name = tensor<string, []>("expand_dims_2_to_fp16"), val = tensor<fp16, [201, 1, 400]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(160960)))];
+            tensor<fp16, [1, 201, 3001]> conv_1_cast_fp16 = conv(dilations = conv_1_dilations_0, groups = conv_1_groups_0, pad = conv_1_pad_0, pad_type = conv_1_pad_type_0, strides = expand_dims_3, weight = expand_dims_2_to_fp16, x = expand_dims_4_cast_fp16)[name = tensor<string, []>("conv_1_cast_fp16")];
+            tensor<int32, [1]> squeeze_0_axes_0 = const()[name = tensor<string, []>("squeeze_0_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [201, 3001]> squeeze_0_cast_fp16 = squeeze(axes = squeeze_0_axes_0, x = conv_0_cast_fp16)[name = tensor<string, []>("squeeze_0_cast_fp16")];
+            tensor<int32, [1]> squeeze_1_axes_0 = const()[name = tensor<string, []>("squeeze_1_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [201, 3001]> squeeze_1_cast_fp16 = squeeze(axes = squeeze_1_axes_0, x = conv_1_cast_fp16)[name = tensor<string, []>("squeeze_1_cast_fp16")];
+            tensor<fp16, [201, 3001]> square_0_cast_fp16 = square(x = squeeze_0_cast_fp16)[name = tensor<string, []>("square_0_cast_fp16")];
+            tensor<fp16, [201, 3001]> square_1_cast_fp16 = square(x = squeeze_1_cast_fp16)[name = tensor<string, []>("square_1_cast_fp16")];
+            tensor<fp16, [201, 3001]> add_1_cast_fp16 = add(x = square_0_cast_fp16, y = square_1_cast_fp16)[name = tensor<string, []>("add_1_cast_fp16")];
+            tensor<fp16, [201, 3001]> magnitudes_1_cast_fp16 = identity(x = add_1_cast_fp16)[name = tensor<string, []>("magnitudes_1_cast_fp16")];
+            tensor<int32, [2]> magnitudes_begin_0 = const()[name = tensor<string, []>("magnitudes_begin_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [2]> magnitudes_end_0 = const()[name = tensor<string, []>("magnitudes_end_0"), val = tensor<int32, [2]>([201, 3000])];
+            tensor<bool, [2]> magnitudes_end_mask_0 = const()[name = tensor<string, []>("magnitudes_end_mask_0"), val = tensor<bool, [2]>([true, false])];
+            tensor<fp16, [201, 3000]> magnitudes_cast_fp16 = slice_by_index(begin = magnitudes_begin_0, end = magnitudes_end_0, end_mask = magnitudes_end_mask_0, x = magnitudes_1_cast_fp16)[name = tensor<string, []>("magnitudes_cast_fp16")];
+            tensor<bool, []> mel_spec_1_transpose_x_0 = const()[name = tensor<string, []>("mel_spec_1_transpose_x_0"), val = tensor<bool, []>(false)];
+            tensor<bool, []> mel_spec_1_transpose_y_0 = const()[name = tensor<string, []>("mel_spec_1_transpose_y_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, [128, 201]> mel_filters_to_fp16 = const()[name = tensor<string, []>("mel_filters_to_fp16"), val = tensor<fp16, [128, 201]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(321856)))];
+            tensor<fp16, [128, 3000]> mel_spec_1_cast_fp16 = matmul(transpose_x = mel_spec_1_transpose_x_0, transpose_y = mel_spec_1_transpose_y_0, x = mel_filters_to_fp16, y = magnitudes_cast_fp16)[name = tensor<string, []>("mel_spec_1_cast_fp16")];
+            tensor<fp16, []> var_41_to_fp16 = const()[name = tensor<string, []>("op_41_to_fp16"), val = tensor<fp16, []>(0x1p-24)];
+            tensor<fp16, [128, 3000]> mel_spec_cast_fp16 = add(x = mel_spec_1_cast_fp16, y = var_41_to_fp16)[name = tensor<string, []>("mel_spec_cast_fp16")];
+            tensor<fp16, []> log_0_epsilon_0_to_fp16 = const()[name = tensor<string, []>("log_0_epsilon_0_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
+            tensor<fp16, [128, 3000]> log_0_cast_fp16 = log(epsilon = log_0_epsilon_0_to_fp16, x = mel_spec_cast_fp16)[name = tensor<string, []>("log_0_cast_fp16")];
+            tensor<fp16, []> mul_0_y_0_to_fp16 = const()[name = tensor<string, []>("mul_0_y_0_to_fp16"), val = tensor<fp16, []>(0x1.bccp-2)];
+            tensor<fp16, [128, 3000]> mul_0_cast_fp16 = mul(x = log_0_cast_fp16, y = mul_0_y_0_to_fp16)[name = tensor<string, []>("mul_0_cast_fp16")];
+            tensor<bool, []> var_44_keep_dims_0 = const()[name = tensor<string, []>("op_44_keep_dims_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, []> var_44_cast_fp16 = reduce_max(keep_dims = var_44_keep_dims_0, x = mul_0_cast_fp16)[name = tensor<string, []>("op_44_cast_fp16")];
+            tensor<fp16, []> var_46_to_fp16 = const()[name = tensor<string, []>("op_46_to_fp16"), val = tensor<fp16, []>(0x1p+3)];
+            tensor<fp16, []> var_47_cast_fp16 = sub(x = var_44_cast_fp16, y = var_46_to_fp16)[name = tensor<string, []>("op_47_cast_fp16")];
+            tensor<fp16, [128, 3000]> log_spec_3_cast_fp16 = maximum(x = mul_0_cast_fp16, y = var_47_cast_fp16)[name = tensor<string, []>("log_spec_3_cast_fp16")];
+            tensor<fp16, []> var_50_to_fp16 = const()[name = tensor<string, []>("op_50_to_fp16"), val = tensor<fp16, []>(0x1p+2)];
+            tensor<fp16, [128, 3000]> var_51_cast_fp16 = add(x = log_spec_3_cast_fp16, y = var_50_to_fp16)[name = tensor<string, []>("op_51_cast_fp16")];
+            tensor<fp16, []> _inversed_log_spec_y_0_to_fp16 = const()[name = tensor<string, []>("_inversed_log_spec_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-2)];
+            tensor<fp16, [128, 3000]> _inversed_log_spec_cast_fp16 = mul(x = var_51_cast_fp16, y = _inversed_log_spec_y_0_to_fp16)[name = tensor<string, []>("_inversed_log_spec_cast_fp16")];
+            tensor<int32, [1]> var_55_axes_0 = const()[name = tensor<string, []>("op_55_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [1, 128, 3000]> var_55_cast_fp16 = expand_dims(axes = var_55_axes_0, x = _inversed_log_spec_cast_fp16)[name = tensor<string, []>("op_55_cast_fp16")];
+            tensor<int32, [1]> var_62_axes_0 = const()[name = tensor<string, []>("op_62_axes_0"), val = tensor<int32, [1]>([2])];
+            tensor<fp16, [1, 128, 1, 3000]> melspectrogram_features = expand_dims(axes = var_62_axes_0, x = var_55_cast_fp16)[name = tensor<string, []>("op_62_cast_fp16")];
+        } -> (melspectrogram_features);
+}

medium/MelSpectrogram.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:009d9fb8f6b589accfa08cebf1c712ef07c3405229ce3cfb3a57ee033c9d8a49
+size 373376

medium/TextDecoder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d00c58a5bc9e64c3f466ce5697c1df117006d67096917d36987a00d03f965e7e
+size 243

medium/TextDecoder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d2e36bbe66f1a4977d0965bce7c797fea3dc14250e96e03918f668669b55d49
+size 694

medium/TextDecoder.mlmodelc/metadata.json

@@ -0,0 +1,171 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Palettized (8 bits), Sparse)",
+    "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,
+      "Squeeze" : 1,
+      "Ios16.gather" : 3,
+      "Ios16.softmax" : 8,
+      "Ios16.reduceMean" : 1,
+      "Split" : 2,
+      "Ios16.linear" : 1,
+      "Ios16.add" : 66,
+      "Concat" : 3,
+      "ExpandDims" : 6,
+      "Ios16.sub" : 1,
+      "Ios16.conv" : 80,
+      "Ios16.gelu" : 4,
+      "Ios16.constexprLutToDense" : 40,
+      "Ios16.constexprSparseToDense" : 41,
+      "Ios16.layerNorm" : 13,
+      "SliceByIndex" : 12,
+      "Ios16.matmul" : 16,
+      "Ios16.batchNorm" : 13,
+      "Ios16.reshape" : 32,
+      "Ios16.mul" : 24
+    },
+    "computePrecision" : "Mixed (Float16, Int32)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "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.conversion_date" : "2026-03-04",
+      "com.github.apple.coremltools.source" : "torch==2.5.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "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_8_bit",
+    "method" : "predict"
+  }
+]

medium/TextDecoder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24eb357578598468c1b847977551748582a0b3d912ffbe65adeb79790582b329
+size 253995124

medium/config.json

@@ -0,0 +1,49 @@
+{
+  "activation_dropout": 0.0,
+  "activation_function": "gelu",
+  "apply_spec_augment": false,
+  "architectures": [
+    "WhisperForConditionalGeneration"
+  ],
+  "attention_dropout": 0.0,
+  "begin_suppress_tokens": [
+    220,
+    50256
+  ],
+  "bos_token_id": 50257,
+  "classifier_proj_size": 256,
+  "d_model": 1280,
+  "decoder_attention_heads": 20,
+  "decoder_ffn_dim": 5120,
+  "decoder_layerdrop": 0.0,
+  "decoder_layers": 4,
+  "decoder_start_token_id": 50258,
+  "dropout": 0.0,
+  "encoder_attention_heads": 20,
+  "encoder_ffn_dim": 5120,
+  "encoder_layerdrop": 0.0,
+  "encoder_layers": 32,
+  "eos_token_id": 50257,
+  "forced_decoder_ids": null,
+  "init_std": 0.02,
+  "is_encoder_decoder": true,
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.05,
+  "max_source_positions": 1500,
+  "max_target_positions": 448,
+  "median_filter_width": 7,
+  "model_type": "whisper",
+  "num_hidden_layers": 32,
+  "num_mel_bins": 128,
+  "pad_token_id": 50257,
+  "scale_embedding": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.51.1",
+  "use_cache": false,
+  "use_weighted_layer_sum": false,
+  "vocab_size": 51866
+}

medium/generation_config.json

@@ -0,0 +1,161 @@
+{
+  "alignment_heads": [
+    [
+      2,
+      4
+    ],
+    [
+      2,
+      11
+    ],
+    [
+      3,
+      3
+    ],
+    [
+      3,
+      6
+    ],
+    [
+      3,
+      11
+    ],
+    [
+      3,
+      14
+    ]
+  ],
+  "attn_implementation": "sdpa",
+  "begin_suppress_tokens": [
+    220,
+    50257
+  ],
+  "bos_token_id": 50257,
+  "decoder_start_token_id": 50258,
+  "eos_token_id": 50257,
+  "forced_decoder_ids": [
+    [
+      1,
+      null
+    ],
+    [
+      2,
+      50360
+    ]
+  ],
+  "is_multilingual": true,
+  "lang_to_id": {
+    "<|af|>": 50327,
+    "<|am|>": 50334,
+    "<|ar|>": 50272,
+    "<|as|>": 50350,
+    "<|az|>": 50304,
+    "<|ba|>": 50355,
+    "<|be|>": 50330,
+    "<|bg|>": 50292,
+    "<|bn|>": 50302,
+    "<|bo|>": 50347,
+    "<|br|>": 50309,
+    "<|bs|>": 50315,
+    "<|ca|>": 50270,
+    "<|cs|>": 50283,
+    "<|cy|>": 50297,
+    "<|da|>": 50285,
+    "<|de|>": 50261,
+    "<|el|>": 50281,
+    "<|en|>": 50259,
+    "<|es|>": 50262,
+    "<|et|>": 50307,
+    "<|eu|>": 50310,
+    "<|fa|>": 50300,
+    "<|fi|>": 50277,
+    "<|fo|>": 50338,
+    "<|fr|>": 50265,
+    "<|gl|>": 50319,
+    "<|gu|>": 50333,
+    "<|haw|>": 50352,
+    "<|ha|>": 50354,
+    "<|he|>": 50279,
+    "<|hi|>": 50276,
+    "<|hr|>": 50291,
+    "<|ht|>": 50339,
+    "<|hu|>": 50286,
+    "<|hy|>": 50312,
+    "<|id|>": 50275,
+    "<|is|>": 50311,
+    "<|it|>": 50274,
+    "<|ja|>": 50266,
+    "<|jw|>": 50356,
+    "<|ka|>": 50329,
+    "<|kk|>": 50316,
+    "<|km|>": 50323,
+    "<|kn|>": 50306,
+    "<|ko|>": 50264,
+    "<|la|>": 50294,
+    "<|lb|>": 50345,
+    "<|ln|>": 50353,
+    "<|lo|>": 50336,
+    "<|lt|>": 50293,
+    "<|lv|>": 50301,
+    "<|mg|>": 50349,
+    "<|mi|>": 50295,
+    "<|mk|>": 50308,
+    "<|ml|>": 50296,
+    "<|mn|>": 50314,
+    "<|mr|>": 50320,
+    "<|ms|>": 50282,
+    "<|mt|>": 50343,
+    "<|my|>": 50346,
+    "<|ne|>": 50313,
+    "<|nl|>": 50271,
+    "<|nn|>": 50342,
+    "<|no|>": 50288,
+    "<|oc|>": 50328,
+    "<|pa|>": 50321,
+    "<|pl|>": 50269,
+    "<|ps|>": 50340,
+    "<|pt|>": 50267,
+    "<|ro|>": 50284,
+    "<|ru|>": 50263,
+    "<|sa|>": 50344,
+    "<|sd|>": 50332,
+    "<|si|>": 50322,
+    "<|sk|>": 50298,
+    "<|sl|>": 50305,
+    "<|sn|>": 50324,
+    "<|so|>": 50326,
+    "<|sq|>": 50317,
+    "<|sr|>": 50303,
+    "<|su|>": 50357,
+    "<|sv|>": 50273,
+    "<|sw|>": 50318,
+    "<|ta|>": 50287,
+    "<|te|>": 50299,
+    "<|tg|>": 50331,
+    "<|th|>": 50289,
+    "<|tk|>": 50341,
+    "<|tl|>": 50348,
+    "<|tr|>": 50268,
+    "<|tt|>": 50351,
+    "<|uk|>": 50280,
+    "<|ur|>": 50290,
+    "<|uz|>": 50337,
+    "<|vi|>": 50278,
+    "<|yi|>": 50335,
+    "<|yo|>": 50325,
+    "<|yue|>": 50358,
+    "<|zh|>": 50260
+  },
+  "max_initial_timestamp_index": 50,
+  "max_length": 448,
+  "no_timestamps_token_id": 50364,
+  "pad_token_id": 50257,
+  "prev_sot_token_id": 50362,
+  "return_timestamps": false,
+  "suppress_tokens": [],
+  "task_to_id": {
+    "transcribe": 50360,
+    "translate": 50359
+  },
+  "transformers_version": "4.51.1"
+}

small/AudioEncoder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:008ccf465f33b6f52e7c1e5edc02b30cc06f80ac33cfc35f8ecd7168e11f186b
+size 243

small/AudioEncoder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f784a1abe9f56ebc42a3ad94536b20f9c4e6f3a41f5705b4d81d06b7c097454
+size 409

small/AudioEncoder.mlmodelc/metadata.json

@@ -0,0 +1,73 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Palettized (4 bits), Sparse)",
+    "outputSchema" : [
+      {
+        "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"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 7,
+    "mlProgramOperationTypeHistogram" : {
+      "Concat" : 672,
+      "Ios16.mul" : 3840,
+      "Ios16.layerNorm" : 65,
+      "SliceByIndex" : 5760,
+      "Ios16.constexprLutToDense" : 194,
+      "Transpose" : 32,
+      "Ios16.einsum" : 7680,
+      "Ios16.conv" : 388,
+      "Ios16.add" : 259,
+      "Ios16.constexprSparseToDense" : 192,
+      "Ios16.softmax" : 3840,
+      "Ios16.gelu" : 34,
+      "Ios16.batchNorm" : 65
+    },
+    "computePrecision" : "Mixed (Float16, Int32)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "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.conversion_date" : "2026-03-05",
+      "com.github.apple.coremltools.source" : "torch==2.5.0",
+      "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 × 128 × 1 × 3000)",
+        "shortDescription" : "",
+        "shape" : "[1, 128, 1, 3000]",
+        "name" : "melspectrogram_features",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "AudioEncoder_mixedBitPalettized_4_bit",
+    "method" : "predict"
+  }
+]

small/AudioEncoder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8326c81c394fa31d6731324964f47b6843f1b8da6c050a85dbb505af0c7d73da
+size 421928256

small/MelSpectrogram.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73b090b1ed51460119344c50a187f8bbbb2d318fccad4ef8bc04280471e03200
+size 243

small/MelSpectrogram.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f0a6e03066f401634997882fc0c48324e31ce40ff1d2a0527d88036af61795f
+size 390

small/MelSpectrogram.mlmodelc/metadata.json

@@ -0,0 +1,75 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Float16",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 128 × 1 × 3000)",
+        "shortDescription" : "",
+        "shape" : "[1, 128, 1, 3000]",
+        "name" : "melspectrogram_features",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 7,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios16.reshape" : 2,
+      "Ios16.mul" : 2,
+      "SliceByIndex" : 1,
+      "Ios16.sub" : 1,
+      "Ios16.log" : 1,
+      "Ios16.square" : 2,
+      "Ios16.add" : 3,
+      "Squeeze" : 2,
+      "Ios16.matmul" : 1,
+      "Ios16.conv" : 2,
+      "Ios16.maximum" : 1,
+      "ExpandDims" : 4,
+      "Ios16.reduceMax" : 1,
+      "Identity" : 1,
+      "Pad" : 1
+    },
+    "computePrecision" : "Mixed (Float16, Int32)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "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.conversion_date" : "2026-03-05",
+      "com.github.apple.coremltools.source" : "torch==2.5.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 480000)",
+        "shortDescription" : "",
+        "shape" : "[480000]",
+        "name" : "audio",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "MelSpectrogram",
+    "method" : "predict"
+  }
+]

small/MelSpectrogram.mlmodelc/model.mil

@@ -0,0 +1,66 @@
+program(1.0)
+[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "3500.14.1"}, {"coremlc-version", "3500.32.1"}, {"coremltools-component-torch", "2.5.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "9.0"}})]
+{
+    func main<ios16>(tensor<fp16, [480000]> audio) {
+            tensor<int32, [3]> var_10 = const()[name = tensor<string, []>("op_10"), val = tensor<int32, [3]>([1, 1, 480000])];
+            tensor<fp16, [1, 1, 480000]> input_1_cast_fp16 = reshape(shape = var_10, x = audio)[name = tensor<string, []>("input_1_cast_fp16")];
+            tensor<int32, [6]> input_3_pad_0 = const()[name = tensor<string, []>("input_3_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 200, 200])];
+            tensor<string, []> input_3_mode_0 = const()[name = tensor<string, []>("input_3_mode_0"), val = tensor<string, []>("reflect")];
+            tensor<fp16, []> const_1_to_fp16 = const()[name = tensor<string, []>("const_1_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
+            tensor<fp16, [1, 1, 480400]> input_3_cast_fp16 = pad(constant_val = const_1_to_fp16, mode = input_3_mode_0, pad = input_3_pad_0, x = input_1_cast_fp16)[name = tensor<string, []>("input_3_cast_fp16")];
+            tensor<int32, [1]> var_22 = const()[name = tensor<string, []>("op_22"), val = tensor<int32, [1]>([480400])];
+            tensor<fp16, [480400]> input_cast_fp16 = reshape(shape = var_22, x = input_3_cast_fp16)[name = tensor<string, []>("input_cast_fp16")];
+            tensor<int32, [1]> expand_dims_0_axes_0 = const()[name = tensor<string, []>("expand_dims_0_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [1, 480400]> expand_dims_0_cast_fp16 = expand_dims(axes = expand_dims_0_axes_0, x = input_cast_fp16)[name = tensor<string, []>("expand_dims_0_cast_fp16")];
+            tensor<int32, [1]> expand_dims_3 = const()[name = tensor<string, []>("expand_dims_3"), val = tensor<int32, [1]>([160])];
+            tensor<int32, [1]> expand_dims_4_axes_0 = const()[name = tensor<string, []>("expand_dims_4_axes_0"), val = tensor<int32, [1]>([1])];
+            tensor<fp16, [1, 1, 480400]> expand_dims_4_cast_fp16 = expand_dims(axes = expand_dims_4_axes_0, x = expand_dims_0_cast_fp16)[name = tensor<string, []>("expand_dims_4_cast_fp16")];
+            tensor<string, []> conv_0_pad_type_0 = const()[name = tensor<string, []>("conv_0_pad_type_0"), val = tensor<string, []>("valid")];
+            tensor<int32, [2]> conv_0_pad_0 = const()[name = tensor<string, []>("conv_0_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> conv_0_dilations_0 = const()[name = tensor<string, []>("conv_0_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, []> conv_0_groups_0 = const()[name = tensor<string, []>("conv_0_groups_0"), val = tensor<int32, []>(1)];
+            tensor<fp16, [201, 1, 400]> expand_dims_1_to_fp16 = const()[name = tensor<string, []>("expand_dims_1_to_fp16"), val = tensor<fp16, [201, 1, 400]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
+            tensor<fp16, [1, 201, 3001]> conv_0_cast_fp16 = conv(dilations = conv_0_dilations_0, groups = conv_0_groups_0, pad = conv_0_pad_0, pad_type = conv_0_pad_type_0, strides = expand_dims_3, weight = expand_dims_1_to_fp16, x = expand_dims_4_cast_fp16)[name = tensor<string, []>("conv_0_cast_fp16")];
+            tensor<string, []> conv_1_pad_type_0 = const()[name = tensor<string, []>("conv_1_pad_type_0"), val = tensor<string, []>("valid")];
+            tensor<int32, [2]> conv_1_pad_0 = const()[name = tensor<string, []>("conv_1_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> conv_1_dilations_0 = const()[name = tensor<string, []>("conv_1_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, []> conv_1_groups_0 = const()[name = tensor<string, []>("conv_1_groups_0"), val = tensor<int32, []>(1)];
+            tensor<fp16, [201, 1, 400]> expand_dims_2_to_fp16 = const()[name = tensor<string, []>("expand_dims_2_to_fp16"), val = tensor<fp16, [201, 1, 400]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(160960)))];
+            tensor<fp16, [1, 201, 3001]> conv_1_cast_fp16 = conv(dilations = conv_1_dilations_0, groups = conv_1_groups_0, pad = conv_1_pad_0, pad_type = conv_1_pad_type_0, strides = expand_dims_3, weight = expand_dims_2_to_fp16, x = expand_dims_4_cast_fp16)[name = tensor<string, []>("conv_1_cast_fp16")];
+            tensor<int32, [1]> squeeze_0_axes_0 = const()[name = tensor<string, []>("squeeze_0_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [201, 3001]> squeeze_0_cast_fp16 = squeeze(axes = squeeze_0_axes_0, x = conv_0_cast_fp16)[name = tensor<string, []>("squeeze_0_cast_fp16")];
+            tensor<int32, [1]> squeeze_1_axes_0 = const()[name = tensor<string, []>("squeeze_1_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [201, 3001]> squeeze_1_cast_fp16 = squeeze(axes = squeeze_1_axes_0, x = conv_1_cast_fp16)[name = tensor<string, []>("squeeze_1_cast_fp16")];
+            tensor<fp16, [201, 3001]> square_0_cast_fp16 = square(x = squeeze_0_cast_fp16)[name = tensor<string, []>("square_0_cast_fp16")];
+            tensor<fp16, [201, 3001]> square_1_cast_fp16 = square(x = squeeze_1_cast_fp16)[name = tensor<string, []>("square_1_cast_fp16")];
+            tensor<fp16, [201, 3001]> add_1_cast_fp16 = add(x = square_0_cast_fp16, y = square_1_cast_fp16)[name = tensor<string, []>("add_1_cast_fp16")];
+            tensor<fp16, [201, 3001]> magnitudes_1_cast_fp16 = identity(x = add_1_cast_fp16)[name = tensor<string, []>("magnitudes_1_cast_fp16")];
+            tensor<int32, [2]> magnitudes_begin_0 = const()[name = tensor<string, []>("magnitudes_begin_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [2]> magnitudes_end_0 = const()[name = tensor<string, []>("magnitudes_end_0"), val = tensor<int32, [2]>([201, 3000])];
+            tensor<bool, [2]> magnitudes_end_mask_0 = const()[name = tensor<string, []>("magnitudes_end_mask_0"), val = tensor<bool, [2]>([true, false])];
+            tensor<fp16, [201, 3000]> magnitudes_cast_fp16 = slice_by_index(begin = magnitudes_begin_0, end = magnitudes_end_0, end_mask = magnitudes_end_mask_0, x = magnitudes_1_cast_fp16)[name = tensor<string, []>("magnitudes_cast_fp16")];
+            tensor<bool, []> mel_spec_1_transpose_x_0 = const()[name = tensor<string, []>("mel_spec_1_transpose_x_0"), val = tensor<bool, []>(false)];
+            tensor<bool, []> mel_spec_1_transpose_y_0 = const()[name = tensor<string, []>("mel_spec_1_transpose_y_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, [128, 201]> mel_filters_to_fp16 = const()[name = tensor<string, []>("mel_filters_to_fp16"), val = tensor<fp16, [128, 201]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(321856)))];
+            tensor<fp16, [128, 3000]> mel_spec_1_cast_fp16 = matmul(transpose_x = mel_spec_1_transpose_x_0, transpose_y = mel_spec_1_transpose_y_0, x = mel_filters_to_fp16, y = magnitudes_cast_fp16)[name = tensor<string, []>("mel_spec_1_cast_fp16")];
+            tensor<fp16, []> var_41_to_fp16 = const()[name = tensor<string, []>("op_41_to_fp16"), val = tensor<fp16, []>(0x1p-24)];
+            tensor<fp16, [128, 3000]> mel_spec_cast_fp16 = add(x = mel_spec_1_cast_fp16, y = var_41_to_fp16)[name = tensor<string, []>("mel_spec_cast_fp16")];
+            tensor<fp16, []> log_0_epsilon_0_to_fp16 = const()[name = tensor<string, []>("log_0_epsilon_0_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
+            tensor<fp16, [128, 3000]> log_0_cast_fp16 = log(epsilon = log_0_epsilon_0_to_fp16, x = mel_spec_cast_fp16)[name = tensor<string, []>("log_0_cast_fp16")];
+            tensor<fp16, []> mul_0_y_0_to_fp16 = const()[name = tensor<string, []>("mul_0_y_0_to_fp16"), val = tensor<fp16, []>(0x1.bccp-2)];
+            tensor<fp16, [128, 3000]> mul_0_cast_fp16 = mul(x = log_0_cast_fp16, y = mul_0_y_0_to_fp16)[name = tensor<string, []>("mul_0_cast_fp16")];
+            tensor<bool, []> var_44_keep_dims_0 = const()[name = tensor<string, []>("op_44_keep_dims_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, []> var_44_cast_fp16 = reduce_max(keep_dims = var_44_keep_dims_0, x = mul_0_cast_fp16)[name = tensor<string, []>("op_44_cast_fp16")];
+            tensor<fp16, []> var_46_to_fp16 = const()[name = tensor<string, []>("op_46_to_fp16"), val = tensor<fp16, []>(0x1p+3)];
+            tensor<fp16, []> var_47_cast_fp16 = sub(x = var_44_cast_fp16, y = var_46_to_fp16)[name = tensor<string, []>("op_47_cast_fp16")];
+            tensor<fp16, [128, 3000]> log_spec_3_cast_fp16 = maximum(x = mul_0_cast_fp16, y = var_47_cast_fp16)[name = tensor<string, []>("log_spec_3_cast_fp16")];
+            tensor<fp16, []> var_50_to_fp16 = const()[name = tensor<string, []>("op_50_to_fp16"), val = tensor<fp16, []>(0x1p+2)];
+            tensor<fp16, [128, 3000]> var_51_cast_fp16 = add(x = log_spec_3_cast_fp16, y = var_50_to_fp16)[name = tensor<string, []>("op_51_cast_fp16")];
+            tensor<fp16, []> _inversed_log_spec_y_0_to_fp16 = const()[name = tensor<string, []>("_inversed_log_spec_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-2)];
+            tensor<fp16, [128, 3000]> _inversed_log_spec_cast_fp16 = mul(x = var_51_cast_fp16, y = _inversed_log_spec_y_0_to_fp16)[name = tensor<string, []>("_inversed_log_spec_cast_fp16")];
+            tensor<int32, [1]> var_55_axes_0 = const()[name = tensor<string, []>("op_55_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [1, 128, 3000]> var_55_cast_fp16 = expand_dims(axes = var_55_axes_0, x = _inversed_log_spec_cast_fp16)[name = tensor<string, []>("op_55_cast_fp16")];
+            tensor<int32, [1]> var_62_axes_0 = const()[name = tensor<string, []>("op_62_axes_0"), val = tensor<int32, [1]>([2])];
+            tensor<fp16, [1, 128, 1, 3000]> melspectrogram_features = expand_dims(axes = var_62_axes_0, x = var_55_cast_fp16)[name = tensor<string, []>("op_62_cast_fp16")];
+        } -> (melspectrogram_features);
+}

small/MelSpectrogram.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:009d9fb8f6b589accfa08cebf1c712ef07c3405229ce3cfb3a57ee033c9d8a49
+size 373376

small/TextDecoder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d00c58a5bc9e64c3f466ce5697c1df117006d67096917d36987a00d03f965e7e
+size 243

small/TextDecoder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d2e36bbe66f1a4977d0965bce7c797fea3dc14250e96e03918f668669b55d49
+size 694

small/TextDecoder.mlmodelc/metadata.json

@@ -0,0 +1,171 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Palettized (8 bits), Sparse)",
+    "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,
+      "Squeeze" : 1,
+      "Ios16.gather" : 3,
+      "Ios16.softmax" : 8,
+      "Ios16.reduceMean" : 1,
+      "Split" : 2,
+      "Ios16.linear" : 1,
+      "Ios16.add" : 66,
+      "Concat" : 3,
+      "ExpandDims" : 6,
+      "Ios16.sub" : 1,
+      "Ios16.conv" : 80,
+      "Ios16.gelu" : 4,
+      "Ios16.constexprLutToDense" : 40,
+      "Ios16.constexprSparseToDense" : 41,
+      "Ios16.layerNorm" : 13,
+      "SliceByIndex" : 12,
+      "Ios16.matmul" : 16,
+      "Ios16.batchNorm" : 13,
+      "Ios16.reshape" : 32,
+      "Ios16.mul" : 24
+    },
+    "computePrecision" : "Mixed (Float16, Int32)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "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.conversion_date" : "2026-03-04",
+      "com.github.apple.coremltools.source" : "torch==2.5.0",
+      "com.github.apple.coremltools.version" : "9.0",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "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_8_bit",
+    "method" : "predict"
+  }
+]

small/TextDecoder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24eb357578598468c1b847977551748582a0b3d912ffbe65adeb79790582b329
+size 253995124

small/config.json

@@ -0,0 +1,49 @@
+{
+  "activation_dropout": 0.0,
+  "activation_function": "gelu",
+  "apply_spec_augment": false,
+  "architectures": [
+    "WhisperForConditionalGeneration"
+  ],
+  "attention_dropout": 0.0,
+  "begin_suppress_tokens": [
+    220,
+    50256
+  ],
+  "bos_token_id": 50257,
+  "classifier_proj_size": 256,
+  "d_model": 1280,
+  "decoder_attention_heads": 20,
+  "decoder_ffn_dim": 5120,
+  "decoder_layerdrop": 0.0,
+  "decoder_layers": 4,
+  "decoder_start_token_id": 50258,
+  "dropout": 0.0,
+  "encoder_attention_heads": 20,
+  "encoder_ffn_dim": 5120,
+  "encoder_layerdrop": 0.0,
+  "encoder_layers": 32,
+  "eos_token_id": 50257,
+  "forced_decoder_ids": null,
+  "init_std": 0.02,
+  "is_encoder_decoder": true,
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.05,
+  "max_source_positions": 1500,
+  "max_target_positions": 448,
+  "median_filter_width": 7,
+  "model_type": "whisper",
+  "num_hidden_layers": 32,
+  "num_mel_bins": 128,
+  "pad_token_id": 50257,
+  "scale_embedding": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.51.1",
+  "use_cache": false,
+  "use_weighted_layer_sum": false,
+  "vocab_size": 51866
+}

small/generation_config.json

@@ -0,0 +1,161 @@
+{
+  "alignment_heads": [
+    [
+      2,
+      4
+    ],
+    [
+      2,
+      11
+    ],
+    [
+      3,
+      3
+    ],
+    [
+      3,
+      6
+    ],
+    [
+      3,
+      11
+    ],
+    [
+      3,
+      14
+    ]
+  ],
+  "attn_implementation": "sdpa",
+  "begin_suppress_tokens": [
+    220,
+    50257
+  ],
+  "bos_token_id": 50257,
+  "decoder_start_token_id": 50258,
+  "eos_token_id": 50257,
+  "forced_decoder_ids": [
+    [
+      1,
+      null
+    ],
+    [
+      2,
+      50360
+    ]
+  ],
+  "is_multilingual": true,
+  "lang_to_id": {
+    "<|af|>": 50327,
+    "<|am|>": 50334,
+    "<|ar|>": 50272,
+    "<|as|>": 50350,
+    "<|az|>": 50304,
+    "<|ba|>": 50355,
+    "<|be|>": 50330,
+    "<|bg|>": 50292,
+    "<|bn|>": 50302,
+    "<|bo|>": 50347,
+    "<|br|>": 50309,
+    "<|bs|>": 50315,
+    "<|ca|>": 50270,
+    "<|cs|>": 50283,
+    "<|cy|>": 50297,
+    "<|da|>": 50285,
+    "<|de|>": 50261,
+    "<|el|>": 50281,
+    "<|en|>": 50259,
+    "<|es|>": 50262,
+    "<|et|>": 50307,
+    "<|eu|>": 50310,
+    "<|fa|>": 50300,
+    "<|fi|>": 50277,
+    "<|fo|>": 50338,
+    "<|fr|>": 50265,
+    "<|gl|>": 50319,
+    "<|gu|>": 50333,
+    "<|haw|>": 50352,
+    "<|ha|>": 50354,
+    "<|he|>": 50279,
+    "<|hi|>": 50276,
+    "<|hr|>": 50291,
+    "<|ht|>": 50339,
+    "<|hu|>": 50286,
+    "<|hy|>": 50312,
+    "<|id|>": 50275,
+    "<|is|>": 50311,
+    "<|it|>": 50274,
+    "<|ja|>": 50266,
+    "<|jw|>": 50356,
+    "<|ka|>": 50329,
+    "<|kk|>": 50316,
+    "<|km|>": 50323,
+    "<|kn|>": 50306,
+    "<|ko|>": 50264,
+    "<|la|>": 50294,
+    "<|lb|>": 50345,
+    "<|ln|>": 50353,
+    "<|lo|>": 50336,
+    "<|lt|>": 50293,
+    "<|lv|>": 50301,
+    "<|mg|>": 50349,
+    "<|mi|>": 50295,
+    "<|mk|>": 50308,
+    "<|ml|>": 50296,
+    "<|mn|>": 50314,
+    "<|mr|>": 50320,
+    "<|ms|>": 50282,
+    "<|mt|>": 50343,
+    "<|my|>": 50346,
+    "<|ne|>": 50313,
+    "<|nl|>": 50271,
+    "<|nn|>": 50342,
+    "<|no|>": 50288,
+    "<|oc|>": 50328,
+    "<|pa|>": 50321,
+    "<|pl|>": 50269,
+    "<|ps|>": 50340,
+    "<|pt|>": 50267,
+    "<|ro|>": 50284,
+    "<|ru|>": 50263,
+    "<|sa|>": 50344,
+    "<|sd|>": 50332,
+    "<|si|>": 50322,
+    "<|sk|>": 50298,
+    "<|sl|>": 50305,
+    "<|sn|>": 50324,
+    "<|so|>": 50326,
+    "<|sq|>": 50317,
+    "<|sr|>": 50303,
+    "<|su|>": 50357,
+    "<|sv|>": 50273,
+    "<|sw|>": 50318,
+    "<|ta|>": 50287,
+    "<|te|>": 50299,
+    "<|tg|>": 50331,
+    "<|th|>": 50289,
+    "<|tk|>": 50341,
+    "<|tl|>": 50348,
+    "<|tr|>": 50268,
+    "<|tt|>": 50351,
+    "<|uk|>": 50280,
+    "<|ur|>": 50290,
+    "<|uz|>": 50337,
+    "<|vi|>": 50278,
+    "<|yi|>": 50335,
+    "<|yo|>": 50325,
+    "<|yue|>": 50358,
+    "<|zh|>": 50260
+  },
+  "max_initial_timestamp_index": 50,
+  "max_length": 448,
+  "no_timestamps_token_id": 50364,
+  "pad_token_id": 50257,
+  "prev_sot_token_id": 50362,
+  "return_timestamps": false,
+  "suppress_tokens": [],
+  "task_to_id": {
+    "transcribe": 50360,
+    "translate": 50359
+  },
+  "transformers_version": "4.51.1"
+}