Add TextDecoderContextPrefil

openai_whisper-large-v3_turbo_1049MB/TextDecoderContextPrefill.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a25601f7f75bf3df5d409d7c868951386b9b9baf79c5af0983b9b219a5978d8
+size 243

openai_whisper-large-v3_turbo_1049MB/TextDecoderContextPrefill.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ea5572d29320308c42c5a2a6f35ade13d9aaf8cae5732b70e5c3c7a7e01d65a
+size 382

openai_whisper-large-v3_turbo_1049MB/TextDecoderContextPrefill.mlmodelc/metadata.json

@@ -0,0 +1,82 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Float16",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 40960 × 1 × 3)",
+        "shortDescription" : "",
+        "shape" : "[1, 40960, 1, 3]",
+        "name" : "key_cache_prefill",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 40960 × 1 × 3)",
+        "shortDescription" : "",
+        "shape" : "[1, 40960, 1, 3]",
+        "name" : "value_cache_prefill",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 8,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios17.mul" : 1,
+      "Ios17.cast" : 1,
+      "Ios17.sub" : 1,
+      "Ios17.reshape" : 2,
+      "Ios17.add" : 1,
+      "Ios17.gather" : 2
+    },
+    "computePrecision" : "Mixed (Float16, Int16, Int32)",
+    "isUpdatable" : "0",
+    "availability" : {
+      "macOS" : "14.0",
+      "tvOS" : "17.0",
+      "visionOS" : "1.0",
+      "watchOS" : "10.0",
+      "iOS" : "17.0",
+      "macCatalyst" : "17.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.source_dialect" : "TorchScript",
+      "com.github.apple.coremltools.source" : "torch==2.1.2",
+      "com.github.apple.coremltools.version" : "7.1"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "task",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "language",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "TextDecoderContextPrefill",
+    "method" : "predict"
+  }
+]

openai_whisper-large-v3_turbo_1049MB/TextDecoderContextPrefill.mlmodelc/model.mil

@@ -0,0 +1,27 @@
+program(1.0)
+[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "5.33.5"}, {"coremlc-version", "1877.40.3"}, {"coremltools-component-torch", "2.1.2"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "7.1"}})]
+{
+    func main<ios17>(tensor<int32, [1]> language, tensor<int32, [1]> task) {
+            tensor<int32, []> var_6 = const()[name = tensor<string, []>("op_6"), val = tensor<int32, []>(50259)];
+            tensor<int32, [1]> var_7 = sub(x = language, y = var_6)[name = tensor<string, []>("op_7")];
+            tensor<int32, []> var_8 = const()[name = tensor<string, []>("op_8"), val = tensor<int32, []>(2)];
+            tensor<int32, [1]> var_9 = mul(x = var_7, y = var_8)[name = tensor<string, []>("op_9")];
+            tensor<int32, [1]> input = add(x = var_9, y = task)[name = tensor<string, []>("input")];
+            tensor<int32, []> var_15_axis_0 = const()[name = tensor<string, []>("op_15_axis_0"), val = tensor<int32, []>(0)];
+            tensor<int32, []> var_15_batch_dims_0 = const()[name = tensor<string, []>("op_15_batch_dims_0"), val = tensor<int32, []>(0)];
+            tensor<bool, []> var_15_validate_indices_0 = const()[name = tensor<string, []>("op_15_validate_indices_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, [200, 122880]> key_cache_lut_weight_to_fp16 = const()[name = tensor<string, []>("key_cache_lut_weight_to_fp16"), val = tensor<fp16, [200, 122880]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
+            tensor<string, []> input_to_int16_dtype_0 = const()[name = tensor<string, []>("input_to_int16_dtype_0"), val = tensor<string, []>("int16")];
+            tensor<int16, [1]> cast_6 = cast(dtype = input_to_int16_dtype_0, x = input)[name = tensor<string, []>("cast_6")];
+            tensor<fp16, [1, 122880]> var_15_cast_fp16_cast_int16 = gather(axis = var_15_axis_0, batch_dims = var_15_batch_dims_0, indices = cast_6, validate_indices = var_15_validate_indices_0, x = key_cache_lut_weight_to_fp16)[name = tensor<string, []>("op_15_cast_fp16_cast_int16")];
+            tensor<int32, [4]> var_20 = const()[name = tensor<string, []>("op_20"), val = tensor<int32, [4]>([1, 40960, 1, 3])];
+            tensor<fp16, [1, 40960, 1, 3]> key_cache_prefill = reshape(shape = var_20, x = var_15_cast_fp16_cast_int16)[name = tensor<string, []>("op_21_cast_fp16")];
+            tensor<int32, []> var_25_axis_0 = const()[name = tensor<string, []>("op_25_axis_0"), val = tensor<int32, []>(0)];
+            tensor<int32, []> var_25_batch_dims_0 = const()[name = tensor<string, []>("op_25_batch_dims_0"), val = tensor<int32, []>(0)];
+            tensor<bool, []> var_25_validate_indices_0 = const()[name = tensor<string, []>("op_25_validate_indices_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, [200, 122880]> value_cache_lut_weight_to_fp16 = const()[name = tensor<string, []>("value_cache_lut_weight_to_fp16"), val = tensor<fp16, [200, 122880]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(49152128)))];
+            tensor<fp16, [1, 122880]> var_25_cast_fp16_cast_int16 = gather(axis = var_25_axis_0, batch_dims = var_25_batch_dims_0, indices = cast_6, validate_indices = var_25_validate_indices_0, x = value_cache_lut_weight_to_fp16)[name = tensor<string, []>("op_25_cast_fp16_cast_int16")];
+            tensor<int32, [4]> var_30 = const()[name = tensor<string, []>("op_30"), val = tensor<int32, [4]>([1, 40960, 1, 3])];
+            tensor<fp16, [1, 40960, 1, 3]> value_cache_prefill = reshape(shape = var_30, x = var_25_cast_fp16_cast_int16)[name = tensor<string, []>("op_31_cast_fp16")];
+        } -> (key_cache_prefill, value_cache_prefill);
+}

openai_whisper-large-v3_turbo_1049MB/TextDecoderContextPrefill.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd4457ee28e369c2f1d7858370352085554a1fb6def48db64f4f4ab27c7015bc
+size 98304192

openai_whisper-large-v3_turbo_1307MB/TextDecoderContextPrefill.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a25601f7f75bf3df5d409d7c868951386b9b9baf79c5af0983b9b219a5978d8
+size 243

openai_whisper-large-v3_turbo_1307MB/TextDecoderContextPrefill.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ea5572d29320308c42c5a2a6f35ade13d9aaf8cae5732b70e5c3c7a7e01d65a
+size 382

openai_whisper-large-v3_turbo_1307MB/TextDecoderContextPrefill.mlmodelc/metadata.json

@@ -0,0 +1,82 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Float16",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 40960 × 1 × 3)",
+        "shortDescription" : "",
+        "shape" : "[1, 40960, 1, 3]",
+        "name" : "key_cache_prefill",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 40960 × 1 × 3)",
+        "shortDescription" : "",
+        "shape" : "[1, 40960, 1, 3]",
+        "name" : "value_cache_prefill",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 8,
+    "mlProgramOperationTypeHistogram" : {
+      "Ios17.mul" : 1,
+      "Ios17.cast" : 1,
+      "Ios17.sub" : 1,
+      "Ios17.reshape" : 2,
+      "Ios17.add" : 1,
+      "Ios17.gather" : 2
+    },
+    "computePrecision" : "Mixed (Float16, Int16, Int32)",
+    "isUpdatable" : "0",
+    "availability" : {
+      "macOS" : "14.0",
+      "tvOS" : "17.0",
+      "visionOS" : "1.0",
+      "watchOS" : "10.0",
+      "iOS" : "17.0",
+      "macCatalyst" : "17.0"
+    },
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "userDefinedMetadata" : {
+      "com.github.apple.coremltools.source_dialect" : "TorchScript",
+      "com.github.apple.coremltools.source" : "torch==2.1.2",
+      "com.github.apple.coremltools.version" : "7.1"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "task",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "language",
+        "type" : "MultiArray"
+      }
+    ],
+    "generatedClassName" : "TextDecoderContextPrefill",
+    "method" : "predict"
+  }
+]

openai_whisper-large-v3_turbo_1307MB/TextDecoderContextPrefill.mlmodelc/model.mil

@@ -0,0 +1,27 @@
+program(1.0)
+[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "5.33.5"}, {"coremlc-version", "1877.40.3"}, {"coremltools-component-torch", "2.1.2"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "7.1"}})]
+{
+    func main<ios17>(tensor<int32, [1]> language, tensor<int32, [1]> task) {
+            tensor<int32, []> var_6 = const()[name = tensor<string, []>("op_6"), val = tensor<int32, []>(50259)];
+            tensor<int32, [1]> var_7 = sub(x = language, y = var_6)[name = tensor<string, []>("op_7")];
+            tensor<int32, []> var_8 = const()[name = tensor<string, []>("op_8"), val = tensor<int32, []>(2)];
+            tensor<int32, [1]> var_9 = mul(x = var_7, y = var_8)[name = tensor<string, []>("op_9")];
+            tensor<int32, [1]> input = add(x = var_9, y = task)[name = tensor<string, []>("input")];
+            tensor<int32, []> var_15_axis_0 = const()[name = tensor<string, []>("op_15_axis_0"), val = tensor<int32, []>(0)];
+            tensor<int32, []> var_15_batch_dims_0 = const()[name = tensor<string, []>("op_15_batch_dims_0"), val = tensor<int32, []>(0)];
+            tensor<bool, []> var_15_validate_indices_0 = const()[name = tensor<string, []>("op_15_validate_indices_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, [200, 122880]> key_cache_lut_weight_to_fp16 = const()[name = tensor<string, []>("key_cache_lut_weight_to_fp16"), val = tensor<fp16, [200, 122880]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
+            tensor<string, []> input_to_int16_dtype_0 = const()[name = tensor<string, []>("input_to_int16_dtype_0"), val = tensor<string, []>("int16")];
+            tensor<int16, [1]> cast_6 = cast(dtype = input_to_int16_dtype_0, x = input)[name = tensor<string, []>("cast_6")];
+            tensor<fp16, [1, 122880]> var_15_cast_fp16_cast_int16 = gather(axis = var_15_axis_0, batch_dims = var_15_batch_dims_0, indices = cast_6, validate_indices = var_15_validate_indices_0, x = key_cache_lut_weight_to_fp16)[name = tensor<string, []>("op_15_cast_fp16_cast_int16")];
+            tensor<int32, [4]> var_20 = const()[name = tensor<string, []>("op_20"), val = tensor<int32, [4]>([1, 40960, 1, 3])];
+            tensor<fp16, [1, 40960, 1, 3]> key_cache_prefill = reshape(shape = var_20, x = var_15_cast_fp16_cast_int16)[name = tensor<string, []>("op_21_cast_fp16")];
+            tensor<int32, []> var_25_axis_0 = const()[name = tensor<string, []>("op_25_axis_0"), val = tensor<int32, []>(0)];
+            tensor<int32, []> var_25_batch_dims_0 = const()[name = tensor<string, []>("op_25_batch_dims_0"), val = tensor<int32, []>(0)];
+            tensor<bool, []> var_25_validate_indices_0 = const()[name = tensor<string, []>("op_25_validate_indices_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, [200, 122880]> value_cache_lut_weight_to_fp16 = const()[name = tensor<string, []>("value_cache_lut_weight_to_fp16"), val = tensor<fp16, [200, 122880]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(49152128)))];
+            tensor<fp16, [1, 122880]> var_25_cast_fp16_cast_int16 = gather(axis = var_25_axis_0, batch_dims = var_25_batch_dims_0, indices = cast_6, validate_indices = var_25_validate_indices_0, x = value_cache_lut_weight_to_fp16)[name = tensor<string, []>("op_25_cast_fp16_cast_int16")];
+            tensor<int32, [4]> var_30 = const()[name = tensor<string, []>("op_30"), val = tensor<int32, [4]>([1, 40960, 1, 3])];
+            tensor<fp16, [1, 40960, 1, 3]> value_cache_prefill = reshape(shape = var_30, x = var_25_cast_fp16_cast_int16)[name = tensor<string, []>("op_31_cast_fp16")];
+        } -> (key_cache_prefill, value_cache_prefill);
+}

openai_whisper-large-v3_turbo_1307MB/TextDecoderContextPrefill.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd4457ee28e369c2f1d7858370352085554a1fb6def48db64f4f4ab27c7015bc
+size 98304192