Upload 9 files

CtcHead.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2bb2d397129247a478fce10652d3011e155cf0247e93086bcdc384a754f005d5
+size 243

CtcHead.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14df3ea46b298ac3efd8141b7dcab77acdaf2d0c700827bc1435093744ae206d
+size 488

CtcHead.mlmodelc/metadata.json

@@ -0,0 +1,67 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "shortDescription" : "CTC decoder head for parakeet-tdt-ctc-110m (encoder_dim=512, vocab=1024+1 blank)",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 188 × 1025)",
+        "shortDescription" : "",
+        "shape" : "[1, 188, 1025]",
+        "name" : "ctc_logits",
+        "type" : "MultiArray"
+      }
+    ],
+    "storagePrecision" : "Float16",
+    "modelParameters" : [
+
+    ],
+    "author" : "Fluid Inference",
+    "specificationVersion" : 8,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios17.cast" : 2,
+      "Ios17.conv" : 1,
+      "Ios17.transpose" : 1,
+      "Ios16.softmax" : 1,
+      "Ios17.log" : 1
+    },
+    "computePrecision" : "Mixed (Float16, Float32, Int32)",
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "14.0",
+      "tvOS" : "17.0",
+      "visionOS" : "1.0",
+      "watchOS" : "10.0",
+      "iOS" : "17.0",
+      "macCatalyst" : "17.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float32",
+        "formattedType" : "MultiArray (Float32 1 × 512 × 188)",
+        "shortDescription" : "",
+        "shape" : "[1, 512, 188]",
+        "name" : "encoder_output",
+        "type" : "MultiArray"
+      }
+    ],
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.conversion_date" : "2026-03-28",
+      "com.github.apple.coremltools.source" : "torch==2.7.0",
+      "com.github.apple.coremltools.version" : "9.0b1",
+      "com.github.apple.coremltools.source_dialect" : "TorchScript"
+    },
+    "generatedClassName" : "CtcHead",
+    "method" : "predict"
+  }
+]

CtcHead.mlmodelc/model.mil

@@ -0,0 +1,24 @@
+program(1.0)
+[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "3520.4.1"}, {"coremlc-version", "3520.5.1"}, {"coremltools-component-torch", "2.7.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "9.0b1"}})]
+{
+    func main<ios17>(tensor<fp32, [1, 512, 188]> encoder_output) {
+            tensor<int32, []> var_4 = const()[name = tensor<string, []>("op_4"), val = tensor<int32, []>(-1)];
+            tensor<string, []> var_18_pad_type_0 = const()[name = tensor<string, []>("op_18_pad_type_0"), val = tensor<string, []>("valid")];
+            tensor<int32, [1]> var_18_strides_0 = const()[name = tensor<string, []>("op_18_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> var_18_pad_0 = const()[name = tensor<string, []>("op_18_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> var_18_dilations_0 = const()[name = tensor<string, []>("op_18_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, []> var_18_groups_0 = const()[name = tensor<string, []>("op_18_groups_0"), val = tensor<int32, []>(1)];
+            tensor<string, []> encoder_output_to_fp16_dtype_0 = const()[name = tensor<string, []>("encoder_output_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
+            tensor<fp16, [1025, 512, 1]> module_decoder_layers_0_weight_to_fp16 = const()[name = tensor<string, []>("module_decoder_layers_0_weight_to_fp16"), val = tensor<fp16, [1025, 512, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
+            tensor<fp16, [1025]> module_decoder_layers_0_bias_to_fp16 = const()[name = tensor<string, []>("module_decoder_layers_0_bias_to_fp16"), val = tensor<fp16, [1025]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1049728)))];
+            tensor<fp16, [1, 512, 188]> encoder_output_to_fp16 = cast(dtype = encoder_output_to_fp16_dtype_0, x = encoder_output)[name = tensor<string, []>("cast_1")];
+            tensor<fp16, [1, 1025, 188]> var_18_cast_fp16 = conv(bias = module_decoder_layers_0_bias_to_fp16, dilations = var_18_dilations_0, groups = var_18_groups_0, pad = var_18_pad_0, pad_type = var_18_pad_type_0, strides = var_18_strides_0, weight = module_decoder_layers_0_weight_to_fp16, x = encoder_output_to_fp16)[name = tensor<string, []>("op_18_cast_fp16")];
+            tensor<int32, [3]> input_perm_0 = const()[name = tensor<string, []>("input_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<fp16, [1, 188, 1025]> input_cast_fp16 = transpose(perm = input_perm_0, x = var_18_cast_fp16)[name = tensor<string, []>("transpose_0")];
+            tensor<fp16, [1, 188, 1025]> out_objects_softmax_cast_fp16 = softmax(axis = var_4, x = input_cast_fp16)[name = tensor<string, []>("out_objects_softmax_cast_fp16")];
+            tensor<fp32, []> out_objects_epsilon_0 = const()[name = tensor<string, []>("out_objects_epsilon_0"), val = tensor<fp32, []>(0x1p-149)];
+            tensor<fp16, [1, 188, 1025]> out_objects_cast_fp16 = log(epsilon = out_objects_epsilon_0, x = out_objects_softmax_cast_fp16)[name = tensor<string, []>("out_objects_cast_fp16")];
+            tensor<string, []> out_objects_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("out_objects_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
+            tensor<fp32, [1, 188, 1025]> ctc_logits = cast(dtype = out_objects_cast_fp16_to_fp32_dtype_0, x = out_objects_cast_fp16)[name = tensor<string, []>("cast_0")];
+        } -> (ctc_logits);
+}

CtcHead.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb9bead064427ffcb7529c0e3f378e421b4dde8e6d81447b6d1ca3352ca850e1
+size 1051842

CtcHead.mlpackage/Data/com.apple.CoreML/model.mlmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de3082e4ab3e934567431b370713514d72e6355e284cae0cda3c8e80cad6fe11
+size 3477

CtcHead.mlpackage/Data/com.apple.CoreML/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb9bead064427ffcb7529c0e3f378e421b4dde8e6d81447b6d1ca3352ca850e1
+size 1051842

CtcHead.mlpackage/Manifest.json

@@ -0,0 +1,18 @@
+{
+    "fileFormatVersion": "1.0.0",
+    "itemInfoEntries": {
+        "3FA85FCF-F3EE-4BA4-B9C9-562CE6B08C20": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Specification",
+            "name": "model.mlmodel",
+            "path": "com.apple.CoreML/model.mlmodel"
+        },
+        "AC32BF24-5F07-4CB2-AED5-C6E41D323170": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Weights",
+            "name": "weights",
+            "path": "com.apple.CoreML/weights"
+        }
+    },
+    "rootModelIdentifier": "3FA85FCF-F3EE-4BA4-B9C9-562CE6B08C20"
+}

ctc_head_metadata.json

@@ -0,0 +1,24 @@
+{
+  "model": "parakeet-tdt-ctc-110m-ctc-head",
+  "source": "nvidia/parakeet-tdt_ctc-110m",
+  "encoder_dim": 512,
+  "time_steps": 188,
+  "vocab_size": 1024,
+  "ctc_classes": 1025,
+  "blank_id": 1024,
+  "max_audio_seconds": 15.0,
+  "input": {
+    "encoder_output": [
+      1,
+      512,
+      188
+    ]
+  },
+  "output": {
+    "ctc_logits": [
+      1,
+      188,
+      1025
+    ]
+  }
+}