Add models

ggml-base-encoder.mlmodelc/analytics/coremldata.bin → ggml-base-encoder.mlmodelc.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2441ae34fc7d12946dba7b63379063e856ffc7c3e11ba5f7533efb1450562ca6
-size 207
+oid sha256:cd106a13783431e3555a79dc58250a0d4f76df6c9f071fc20f6dfd3dc84be6ea
+size 37922607

ggml-base-encoder.mlmodelc/metadata.json

@@ -1,64 +0,0 @@
-[
-  {
-    "metadataOutputVersion" : "3.0",
-    "storagePrecision" : "Float16",
-    "outputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32)",
-        "shortDescription" : "",
-        "shape" : "[]",
-        "name" : "output",
-        "type" : "MultiArray"
-      }
-    ],
-    "modelParameters" : [
-
-    ],
-    "specificationVersion" : 6,
-    "mlProgramOperationTypeHistogram" : {
-      "Linear" : 36,
-      "Matmul" : 12,
-      "Cast" : 2,
-      "Conv" : 2,
-      "Softmax" : 6,
-      "Add" : 13,
-      "LayerNorm" : 13,
-      "Mul" : 12,
-      "Transpose" : 25,
-      "Gelu" : 8,
-      "Reshape" : 24
-    },
-    "computePrecision" : "Mixed (Float16, Float32, Int32)",
-    "isUpdatable" : "0",
-    "availability" : {
-      "macOS" : "12.0",
-      "tvOS" : "15.0",
-      "watchOS" : "8.0",
-      "iOS" : "15.0",
-      "macCatalyst" : "15.0"
-    },
-    "modelType" : {
-      "name" : "MLModelType_mlProgram"
-    },
-    "userDefinedMetadata" : {
-
-    },
-    "inputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32 1 × 80 × 3000)",
-        "shortDescription" : "",
-        "shape" : "[1, 80, 3000]",
-        "name" : "logmel_data",
-        "type" : "MultiArray"
-      }
-    ],
-    "generatedClassName" : "coreml_encoder_base",
-    "method" : "predict"
-  }
-]

ggml-base-encoder.mlmodelc/model.mil

@@ -1,393 +0,0 @@
-program(1.0)
-[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "4.28.4"}, {"coremlc-version", "1436.100.10"}})]
-{
-    func main<ios15>(tensor<fp32, [1, 80, 3000]> logmel_data) {
-            tensor<int32, []> var_20 = const()[name = tensor<string, []>("op_20"), val = tensor<int32, []>(1)];
-            tensor<int32, [1]> var_28 = const()[name = tensor<string, []>("op_28"), val = tensor<int32, [1]>([1])];
-            tensor<int32, [1]> var_30 = const()[name = tensor<string, []>("op_30"), val = tensor<int32, [1]>([1])];
-            tensor<string, []> var_32_pad_type_0 = const()[name = tensor<string, []>("op_32_pad_type_0"), val = tensor<string, []>("custom")];
-            tensor<int32, [2]> var_32_pad_0 = const()[name = tensor<string, []>("op_32_pad_0"), val = tensor<int32, [2]>([1, 1])];
-            tensor<string, []> logmel_data_to_fp16_dtype_0 = const()[name = tensor<string, []>("logmel_data_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
-            tensor<fp16, [512, 80, 3]> weight_3_to_fp16 = const()[name = tensor<string, []>("weight_3_to_fp16"), val = tensor<fp16, [512, 80, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
-            tensor<fp16, [512]> bias_3_to_fp16 = const()[name = tensor<string, []>("bias_3_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(245888)))];
-            tensor<fp16, [1, 80, 3000]> cast_187 = cast(dtype = logmel_data_to_fp16_dtype_0, x = logmel_data);
-            tensor<fp16, [1, 512, 3000]> var_32_cast = conv(bias = bias_3_to_fp16, dilations = var_30, groups = var_20, pad = var_32_pad_0, pad_type = var_32_pad_type_0, strides = var_28, weight = weight_3_to_fp16, x = cast_187);
-            tensor<string, []> input_1_mode_0 = const()[name = tensor<string, []>("input_1_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 512, 3000]> input_1_cast = gelu(mode = input_1_mode_0, x = var_32_cast);
-            tensor<int32, []> var_36 = const()[name = tensor<string, []>("op_36"), val = tensor<int32, []>(1)];
-            tensor<int32, [1]> var_45 = const()[name = tensor<string, []>("op_45"), val = tensor<int32, [1]>([2])];
-            tensor<int32, [1]> var_47 = const()[name = tensor<string, []>("op_47"), val = tensor<int32, [1]>([1])];
-            tensor<string, []> var_49_pad_type_0 = const()[name = tensor<string, []>("op_49_pad_type_0"), val = tensor<string, []>("custom")];
-            tensor<int32, [2]> var_49_pad_0 = const()[name = tensor<string, []>("op_49_pad_0"), val = tensor<int32, [2]>([1, 1])];
-            tensor<fp16, [512, 512, 3]> weight_7_to_fp16 = const()[name = tensor<string, []>("weight_7_to_fp16"), val = tensor<fp16, [512, 512, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(246976)))];
-            tensor<fp16, [512]> bias_7_to_fp16 = const()[name = tensor<string, []>("bias_7_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1819904)))];
-            tensor<fp16, [1, 512, 1500]> var_49_cast = conv(bias = bias_7_to_fp16, dilations = var_47, groups = var_36, pad = var_49_pad_0, pad_type = var_49_pad_type_0, strides = var_45, weight = weight_7_to_fp16, x = input_1_cast);
-            tensor<string, []> x_3_mode_0 = const()[name = tensor<string, []>("x_3_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 512, 1500]> x_3_cast = gelu(mode = x_3_mode_0, x = var_49_cast);
-            tensor<int32, [3]> var_54 = const()[name = tensor<string, []>("op_54"), val = tensor<int32, [3]>([0, 2, 1])];
-            tensor<fp16, [1500, 512]> positional_embedding_to_fp16 = const()[name = tensor<string, []>("positional_embedding_to_fp16"), val = tensor<fp16, [1500, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1820992)))];
-            tensor<fp16, [1, 1500, 512]> transpose_48 = transpose(perm = var_54, x = x_3_cast);
-            tensor<fp16, [1, 1500, 512]> var_57_cast = add(x = transpose_48, y = positional_embedding_to_fp16);
-            tensor<int32, []> var_70 = const()[name = tensor<string, []>("op_70"), val = tensor<int32, []>(-1)];
-            tensor<int32, [1]> var_87_axes_0 = const()[name = tensor<string, []>("op_87_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_0_attn_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_0_attn_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3357056)))];
-            tensor<fp16, [512]> blocks_0_attn_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_0_attn_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3358144)))];
-            tensor<fp16, []> var_76_to_fp16 = const()[name = tensor<string, []>("op_76_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 512]> var_87_cast = layer_norm(axes = var_87_axes_0, beta = blocks_0_attn_ln_bias_to_fp16, epsilon = var_76_to_fp16, gamma = blocks_0_attn_ln_weight_to_fp16, x = var_57_cast);
-            tensor<fp16, [512, 512]> var_98_to_fp16 = const()[name = tensor<string, []>("op_98_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3359232)))];
-            tensor<fp16, [512]> var_99_to_fp16 = const()[name = tensor<string, []>("op_99_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3883584)))];
-            tensor<fp16, [1, 1500, 512]> q_1_cast = linear(bias = var_99_to_fp16, weight = var_98_to_fp16, x = var_87_cast);
-            tensor<fp16, [512, 512]> var_102_to_fp16 = const()[name = tensor<string, []>("op_102_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3884672)))];
-            tensor<fp16, [512]> k_1_bias_0_to_fp16 = const()[name = tensor<string, []>("k_1_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4409024)))];
-            tensor<fp16, [1, 1500, 512]> k_1_cast = linear(bias = k_1_bias_0_to_fp16, weight = var_102_to_fp16, x = var_87_cast);
-            tensor<fp16, [512, 512]> var_106_to_fp16 = const()[name = tensor<string, []>("op_106_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4410112)))];
-            tensor<fp16, [512]> var_107_to_fp16 = const()[name = tensor<string, []>("op_107_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4934464)))];
-            tensor<fp16, [1, 1500, 512]> v_1_cast = linear(bias = var_107_to_fp16, weight = var_106_to_fp16, x = var_87_cast);
-            tensor<int32, [4]> var_115 = const()[name = tensor<string, []>("op_115"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_116_cast = reshape(shape = var_115, x = q_1_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_42_to_fp16 = const()[name = tensor<string, []>("const_42_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> q_3_cast = mul(x = var_116_cast, y = const_42_to_fp16);
-            tensor<int32, [4]> var_122 = const()[name = tensor<string, []>("op_122"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_123_cast = reshape(shape = var_122, x = k_1_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_43_to_fp16 = const()[name = tensor<string, []>("const_43_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> k_3_cast = mul(x = var_123_cast, y = const_43_to_fp16);
-            tensor<int32, [4]> var_129 = const()[name = tensor<string, []>("op_129"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_130_cast = reshape(shape = var_129, x = v_1_cast);
-            tensor<int32, [4]> var_131 = const()[name = tensor<string, []>("op_131"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<bool, []> qk_1_transpose_x_0 = const()[name = tensor<string, []>("qk_1_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> qk_1_transpose_y_0 = const()[name = tensor<string, []>("qk_1_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<int32, [4]> transpose_12_perm_0 = const()[name = tensor<string, []>("transpose_12_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [4]> transpose_13_perm_0 = const()[name = tensor<string, []>("transpose_13_perm_0"), val = tensor<int32, [4]>([0, 2, 3, 1])];
-            tensor<fp16, [1, 8, 64, 1500]> transpose_45 = transpose(perm = transpose_13_perm_0, x = k_3_cast);
-            tensor<fp16, [1, 8, 1500, 64]> transpose_46 = transpose(perm = transpose_12_perm_0, x = q_3_cast);
-            tensor<fp16, [1, 8, 1500, 1500]> qk_1_cast = matmul(transpose_x = qk_1_transpose_x_0, transpose_y = qk_1_transpose_y_0, x = transpose_46, y = transpose_45);
-            tensor<fp16, [1, 8, 1500, 1500]> var_135_cast = softmax(axis = var_70, x = qk_1_cast);
-            tensor<bool, []> var_137_transpose_x_0 = const()[name = tensor<string, []>("op_137_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> var_137_transpose_y_0 = const()[name = tensor<string, []>("op_137_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [1, 8, 1500, 64]> transpose_47 = transpose(perm = var_131, x = var_130_cast);
-            tensor<fp16, [1, 8, 1500, 64]> var_137_cast = matmul(transpose_x = var_137_transpose_x_0, transpose_y = var_137_transpose_y_0, x = var_135_cast, y = transpose_47);
-            tensor<int32, [4]> var_138 = const()[name = tensor<string, []>("op_138"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [3]> concat_0 = const()[name = tensor<string, []>("concat_0"), val = tensor<int32, [3]>([1, 1500, 512])];
-            tensor<fp16, [1, 1500, 8, 64]> transpose_44 = transpose(perm = var_138, x = var_137_cast);
-            tensor<fp16, [1, 1500, 512]> x_11_cast = reshape(shape = concat_0, x = transpose_44);
-            tensor<fp16, [512, 512]> var_143_to_fp16 = const()[name = tensor<string, []>("op_143_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4935552)))];
-            tensor<fp16, [512]> var_144_to_fp16 = const()[name = tensor<string, []>("op_144_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5459904)))];
-            tensor<fp16, [1, 1500, 512]> var_145_cast = linear(bias = var_144_to_fp16, weight = var_143_to_fp16, x = x_11_cast);
-            tensor<fp16, [1, 1500, 512]> x_13_cast = add(x = var_57_cast, y = var_145_cast);
-            tensor<int32, [1]> var_151_axes_0 = const()[name = tensor<string, []>("op_151_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_0_mlp_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_0_mlp_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5460992)))];
-            tensor<fp16, [512]> blocks_0_mlp_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_0_mlp_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5462080)))];
-            tensor<fp16, [1, 1500, 512]> var_151_cast = layer_norm(axes = var_151_axes_0, beta = blocks_0_mlp_ln_bias_to_fp16, epsilon = var_76_to_fp16, gamma = blocks_0_mlp_ln_weight_to_fp16, x = x_13_cast);
-            tensor<fp16, [2048, 512]> var_160_to_fp16 = const()[name = tensor<string, []>("op_160_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5463168)))];
-            tensor<fp16, [2048]> var_161_to_fp16 = const()[name = tensor<string, []>("op_161_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(7560384)))];
-            tensor<fp16, [1, 1500, 2048]> input_9_cast = linear(bias = var_161_to_fp16, weight = var_160_to_fp16, x = var_151_cast);
-            tensor<string, []> x_17_mode_0 = const()[name = tensor<string, []>("x_17_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 1500, 2048]> x_17_cast = gelu(mode = x_17_mode_0, x = input_9_cast);
-            tensor<fp16, [512, 2048]> var_166_to_fp16 = const()[name = tensor<string, []>("op_166_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(7564544)))];
-            tensor<fp16, [512]> var_167_to_fp16 = const()[name = tensor<string, []>("op_167_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9661760)))];
-            tensor<fp16, [1, 1500, 512]> var_168_cast = linear(bias = var_167_to_fp16, weight = var_166_to_fp16, x = x_17_cast);
-            tensor<fp16, [1, 1500, 512]> x_19_cast = add(x = x_13_cast, y = var_168_cast);
-            tensor<int32, []> var_177 = const()[name = tensor<string, []>("op_177"), val = tensor<int32, []>(-1)];
-            tensor<int32, [1]> var_194_axes_0 = const()[name = tensor<string, []>("op_194_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_1_attn_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_1_attn_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9662848)))];
-            tensor<fp16, [512]> blocks_1_attn_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_1_attn_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9663936)))];
-            tensor<fp16, []> var_183_to_fp16 = const()[name = tensor<string, []>("op_183_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 512]> var_194_cast = layer_norm(axes = var_194_axes_0, beta = blocks_1_attn_ln_bias_to_fp16, epsilon = var_183_to_fp16, gamma = blocks_1_attn_ln_weight_to_fp16, x = x_19_cast);
-            tensor<fp16, [512, 512]> var_205_to_fp16 = const()[name = tensor<string, []>("op_205_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9665024)))];
-            tensor<fp16, [512]> var_206_to_fp16 = const()[name = tensor<string, []>("op_206_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10189376)))];
-            tensor<fp16, [1, 1500, 512]> q_5_cast = linear(bias = var_206_to_fp16, weight = var_205_to_fp16, x = var_194_cast);
-            tensor<fp16, [512, 512]> var_209_to_fp16 = const()[name = tensor<string, []>("op_209_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10190464)))];
-            tensor<fp16, [512]> k_5_bias_0_to_fp16 = const()[name = tensor<string, []>("k_5_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10714816)))];
-            tensor<fp16, [1, 1500, 512]> k_5_cast = linear(bias = k_5_bias_0_to_fp16, weight = var_209_to_fp16, x = var_194_cast);
-            tensor<fp16, [512, 512]> var_213_to_fp16 = const()[name = tensor<string, []>("op_213_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10715904)))];
-            tensor<fp16, [512]> var_214_to_fp16 = const()[name = tensor<string, []>("op_214_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(11240256)))];
-            tensor<fp16, [1, 1500, 512]> v_5_cast = linear(bias = var_214_to_fp16, weight = var_213_to_fp16, x = var_194_cast);
-            tensor<int32, [4]> var_222 = const()[name = tensor<string, []>("op_222"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_223_cast = reshape(shape = var_222, x = q_5_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_44_to_fp16 = const()[name = tensor<string, []>("const_44_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> q_7_cast = mul(x = var_223_cast, y = const_44_to_fp16);
-            tensor<int32, [4]> var_229 = const()[name = tensor<string, []>("op_229"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_230_cast = reshape(shape = var_229, x = k_5_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_45_to_fp16 = const()[name = tensor<string, []>("const_45_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> k_7_cast = mul(x = var_230_cast, y = const_45_to_fp16);
-            tensor<int32, [4]> var_236 = const()[name = tensor<string, []>("op_236"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_237_cast = reshape(shape = var_236, x = v_5_cast);
-            tensor<int32, [4]> var_238 = const()[name = tensor<string, []>("op_238"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<bool, []> qk_3_transpose_x_0 = const()[name = tensor<string, []>("qk_3_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> qk_3_transpose_y_0 = const()[name = tensor<string, []>("qk_3_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<int32, [4]> transpose_14_perm_0 = const()[name = tensor<string, []>("transpose_14_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [4]> transpose_15_perm_0 = const()[name = tensor<string, []>("transpose_15_perm_0"), val = tensor<int32, [4]>([0, 2, 3, 1])];
-            tensor<fp16, [1, 8, 64, 1500]> transpose_41 = transpose(perm = transpose_15_perm_0, x = k_7_cast);
-            tensor<fp16, [1, 8, 1500, 64]> transpose_42 = transpose(perm = transpose_14_perm_0, x = q_7_cast);
-            tensor<fp16, [1, 8, 1500, 1500]> qk_3_cast = matmul(transpose_x = qk_3_transpose_x_0, transpose_y = qk_3_transpose_y_0, x = transpose_42, y = transpose_41);
-            tensor<fp16, [1, 8, 1500, 1500]> var_242_cast = softmax(axis = var_177, x = qk_3_cast);
-            tensor<bool, []> var_244_transpose_x_0 = const()[name = tensor<string, []>("op_244_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> var_244_transpose_y_0 = const()[name = tensor<string, []>("op_244_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [1, 8, 1500, 64]> transpose_43 = transpose(perm = var_238, x = var_237_cast);
-            tensor<fp16, [1, 8, 1500, 64]> var_244_cast = matmul(transpose_x = var_244_transpose_x_0, transpose_y = var_244_transpose_y_0, x = var_242_cast, y = transpose_43);
-            tensor<int32, [4]> var_245 = const()[name = tensor<string, []>("op_245"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [3]> concat_1 = const()[name = tensor<string, []>("concat_1"), val = tensor<int32, [3]>([1, 1500, 512])];
-            tensor<fp16, [1, 1500, 8, 64]> transpose_40 = transpose(perm = var_245, x = var_244_cast);
-            tensor<fp16, [1, 1500, 512]> x_23_cast = reshape(shape = concat_1, x = transpose_40);
-            tensor<fp16, [512, 512]> var_250_to_fp16 = const()[name = tensor<string, []>("op_250_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(11241344)))];
-            tensor<fp16, [512]> var_251_to_fp16 = const()[name = tensor<string, []>("op_251_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(11765696)))];
-            tensor<fp16, [1, 1500, 512]> var_252_cast = linear(bias = var_251_to_fp16, weight = var_250_to_fp16, x = x_23_cast);
-            tensor<fp16, [1, 1500, 512]> x_25_cast = add(x = x_19_cast, y = var_252_cast);
-            tensor<int32, [1]> var_258_axes_0 = const()[name = tensor<string, []>("op_258_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_1_mlp_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_1_mlp_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(11766784)))];
-            tensor<fp16, [512]> blocks_1_mlp_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_1_mlp_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(11767872)))];
-            tensor<fp16, [1, 1500, 512]> var_258_cast = layer_norm(axes = var_258_axes_0, beta = blocks_1_mlp_ln_bias_to_fp16, epsilon = var_183_to_fp16, gamma = blocks_1_mlp_ln_weight_to_fp16, x = x_25_cast);
-            tensor<fp16, [2048, 512]> var_267_to_fp16 = const()[name = tensor<string, []>("op_267_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(11768960)))];
-            tensor<fp16, [2048]> var_268_to_fp16 = const()[name = tensor<string, []>("op_268_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(13866176)))];
-            tensor<fp16, [1, 1500, 2048]> input_17_cast = linear(bias = var_268_to_fp16, weight = var_267_to_fp16, x = var_258_cast);
-            tensor<string, []> x_29_mode_0 = const()[name = tensor<string, []>("x_29_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 1500, 2048]> x_29_cast = gelu(mode = x_29_mode_0, x = input_17_cast);
-            tensor<fp16, [512, 2048]> var_273_to_fp16 = const()[name = tensor<string, []>("op_273_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(13870336)))];
-            tensor<fp16, [512]> var_274_to_fp16 = const()[name = tensor<string, []>("op_274_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15967552)))];
-            tensor<fp16, [1, 1500, 512]> var_275_cast = linear(bias = var_274_to_fp16, weight = var_273_to_fp16, x = x_29_cast);
-            tensor<fp16, [1, 1500, 512]> x_31_cast = add(x = x_25_cast, y = var_275_cast);
-            tensor<int32, []> var_284 = const()[name = tensor<string, []>("op_284"), val = tensor<int32, []>(-1)];
-            tensor<int32, [1]> var_301_axes_0 = const()[name = tensor<string, []>("op_301_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_2_attn_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_2_attn_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15968640)))];
-            tensor<fp16, [512]> blocks_2_attn_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_2_attn_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15969728)))];
-            tensor<fp16, []> var_290_to_fp16 = const()[name = tensor<string, []>("op_290_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 512]> var_301_cast = layer_norm(axes = var_301_axes_0, beta = blocks_2_attn_ln_bias_to_fp16, epsilon = var_290_to_fp16, gamma = blocks_2_attn_ln_weight_to_fp16, x = x_31_cast);
-            tensor<fp16, [512, 512]> var_312_to_fp16 = const()[name = tensor<string, []>("op_312_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15970816)))];
-            tensor<fp16, [512]> var_313_to_fp16 = const()[name = tensor<string, []>("op_313_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(16495168)))];
-            tensor<fp16, [1, 1500, 512]> q_9_cast = linear(bias = var_313_to_fp16, weight = var_312_to_fp16, x = var_301_cast);
-            tensor<fp16, [512, 512]> var_316_to_fp16 = const()[name = tensor<string, []>("op_316_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(16496256)))];
-            tensor<fp16, [512]> k_9_bias_0_to_fp16 = const()[name = tensor<string, []>("k_9_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(17020608)))];
-            tensor<fp16, [1, 1500, 512]> k_9_cast = linear(bias = k_9_bias_0_to_fp16, weight = var_316_to_fp16, x = var_301_cast);
-            tensor<fp16, [512, 512]> var_320_to_fp16 = const()[name = tensor<string, []>("op_320_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(17021696)))];
-            tensor<fp16, [512]> var_321_to_fp16 = const()[name = tensor<string, []>("op_321_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(17546048)))];
-            tensor<fp16, [1, 1500, 512]> v_9_cast = linear(bias = var_321_to_fp16, weight = var_320_to_fp16, x = var_301_cast);
-            tensor<int32, [4]> var_329 = const()[name = tensor<string, []>("op_329"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_330_cast = reshape(shape = var_329, x = q_9_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_46_to_fp16 = const()[name = tensor<string, []>("const_46_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> q_11_cast = mul(x = var_330_cast, y = const_46_to_fp16);
-            tensor<int32, [4]> var_336 = const()[name = tensor<string, []>("op_336"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_337_cast = reshape(shape = var_336, x = k_9_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_47_to_fp16 = const()[name = tensor<string, []>("const_47_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> k_11_cast = mul(x = var_337_cast, y = const_47_to_fp16);
-            tensor<int32, [4]> var_343 = const()[name = tensor<string, []>("op_343"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_344_cast = reshape(shape = var_343, x = v_9_cast);
-            tensor<int32, [4]> var_345 = const()[name = tensor<string, []>("op_345"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<bool, []> qk_5_transpose_x_0 = const()[name = tensor<string, []>("qk_5_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> qk_5_transpose_y_0 = const()[name = tensor<string, []>("qk_5_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<int32, [4]> transpose_16_perm_0 = const()[name = tensor<string, []>("transpose_16_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [4]> transpose_17_perm_0 = const()[name = tensor<string, []>("transpose_17_perm_0"), val = tensor<int32, [4]>([0, 2, 3, 1])];
-            tensor<fp16, [1, 8, 64, 1500]> transpose_37 = transpose(perm = transpose_17_perm_0, x = k_11_cast);
-            tensor<fp16, [1, 8, 1500, 64]> transpose_38 = transpose(perm = transpose_16_perm_0, x = q_11_cast);
-            tensor<fp16, [1, 8, 1500, 1500]> qk_5_cast = matmul(transpose_x = qk_5_transpose_x_0, transpose_y = qk_5_transpose_y_0, x = transpose_38, y = transpose_37);
-            tensor<fp16, [1, 8, 1500, 1500]> var_349_cast = softmax(axis = var_284, x = qk_5_cast);
-            tensor<bool, []> var_351_transpose_x_0 = const()[name = tensor<string, []>("op_351_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> var_351_transpose_y_0 = const()[name = tensor<string, []>("op_351_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [1, 8, 1500, 64]> transpose_39 = transpose(perm = var_345, x = var_344_cast);
-            tensor<fp16, [1, 8, 1500, 64]> var_351_cast = matmul(transpose_x = var_351_transpose_x_0, transpose_y = var_351_transpose_y_0, x = var_349_cast, y = transpose_39);
-            tensor<int32, [4]> var_352 = const()[name = tensor<string, []>("op_352"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [3]> concat_2 = const()[name = tensor<string, []>("concat_2"), val = tensor<int32, [3]>([1, 1500, 512])];
-            tensor<fp16, [1, 1500, 8, 64]> transpose_36 = transpose(perm = var_352, x = var_351_cast);
-            tensor<fp16, [1, 1500, 512]> x_35_cast = reshape(shape = concat_2, x = transpose_36);
-            tensor<fp16, [512, 512]> var_357_to_fp16 = const()[name = tensor<string, []>("op_357_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(17547136)))];
-            tensor<fp16, [512]> var_358_to_fp16 = const()[name = tensor<string, []>("op_358_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(18071488)))];
-            tensor<fp16, [1, 1500, 512]> var_359_cast = linear(bias = var_358_to_fp16, weight = var_357_to_fp16, x = x_35_cast);
-            tensor<fp16, [1, 1500, 512]> x_37_cast = add(x = x_31_cast, y = var_359_cast);
-            tensor<int32, [1]> var_365_axes_0 = const()[name = tensor<string, []>("op_365_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_2_mlp_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_2_mlp_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(18072576)))];
-            tensor<fp16, [512]> blocks_2_mlp_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_2_mlp_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(18073664)))];
-            tensor<fp16, [1, 1500, 512]> var_365_cast = layer_norm(axes = var_365_axes_0, beta = blocks_2_mlp_ln_bias_to_fp16, epsilon = var_290_to_fp16, gamma = blocks_2_mlp_ln_weight_to_fp16, x = x_37_cast);
-            tensor<fp16, [2048, 512]> var_374_to_fp16 = const()[name = tensor<string, []>("op_374_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(18074752)))];
-            tensor<fp16, [2048]> var_375_to_fp16 = const()[name = tensor<string, []>("op_375_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(20171968)))];
-            tensor<fp16, [1, 1500, 2048]> input_25_cast = linear(bias = var_375_to_fp16, weight = var_374_to_fp16, x = var_365_cast);
-            tensor<string, []> x_41_mode_0 = const()[name = tensor<string, []>("x_41_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 1500, 2048]> x_41_cast = gelu(mode = x_41_mode_0, x = input_25_cast);
-            tensor<fp16, [512, 2048]> var_380_to_fp16 = const()[name = tensor<string, []>("op_380_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(20176128)))];
-            tensor<fp16, [512]> var_381_to_fp16 = const()[name = tensor<string, []>("op_381_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(22273344)))];
-            tensor<fp16, [1, 1500, 512]> var_382_cast = linear(bias = var_381_to_fp16, weight = var_380_to_fp16, x = x_41_cast);
-            tensor<fp16, [1, 1500, 512]> x_43_cast = add(x = x_37_cast, y = var_382_cast);
-            tensor<int32, []> var_391 = const()[name = tensor<string, []>("op_391"), val = tensor<int32, []>(-1)];
-            tensor<int32, [1]> var_408_axes_0 = const()[name = tensor<string, []>("op_408_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_3_attn_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_3_attn_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(22274432)))];
-            tensor<fp16, [512]> blocks_3_attn_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_3_attn_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(22275520)))];
-            tensor<fp16, []> var_397_to_fp16 = const()[name = tensor<string, []>("op_397_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 512]> var_408_cast = layer_norm(axes = var_408_axes_0, beta = blocks_3_attn_ln_bias_to_fp16, epsilon = var_397_to_fp16, gamma = blocks_3_attn_ln_weight_to_fp16, x = x_43_cast);
-            tensor<fp16, [512, 512]> var_419_to_fp16 = const()[name = tensor<string, []>("op_419_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(22276608)))];
-            tensor<fp16, [512]> var_420_to_fp16 = const()[name = tensor<string, []>("op_420_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(22800960)))];
-            tensor<fp16, [1, 1500, 512]> q_13_cast = linear(bias = var_420_to_fp16, weight = var_419_to_fp16, x = var_408_cast);
-            tensor<fp16, [512, 512]> var_423_to_fp16 = const()[name = tensor<string, []>("op_423_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(22802048)))];
-            tensor<fp16, [512]> k_13_bias_0_to_fp16 = const()[name = tensor<string, []>("k_13_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(23326400)))];
-            tensor<fp16, [1, 1500, 512]> k_13_cast = linear(bias = k_13_bias_0_to_fp16, weight = var_423_to_fp16, x = var_408_cast);
-            tensor<fp16, [512, 512]> var_427_to_fp16 = const()[name = tensor<string, []>("op_427_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(23327488)))];
-            tensor<fp16, [512]> var_428_to_fp16 = const()[name = tensor<string, []>("op_428_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(23851840)))];
-            tensor<fp16, [1, 1500, 512]> v_13_cast = linear(bias = var_428_to_fp16, weight = var_427_to_fp16, x = var_408_cast);
-            tensor<int32, [4]> var_436 = const()[name = tensor<string, []>("op_436"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_437_cast = reshape(shape = var_436, x = q_13_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_48_to_fp16 = const()[name = tensor<string, []>("const_48_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> q_15_cast = mul(x = var_437_cast, y = const_48_to_fp16);
-            tensor<int32, [4]> var_443 = const()[name = tensor<string, []>("op_443"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_444_cast = reshape(shape = var_443, x = k_13_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_49_to_fp16 = const()[name = tensor<string, []>("const_49_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> k_15_cast = mul(x = var_444_cast, y = const_49_to_fp16);
-            tensor<int32, [4]> var_450 = const()[name = tensor<string, []>("op_450"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_451_cast = reshape(shape = var_450, x = v_13_cast);
-            tensor<int32, [4]> var_452 = const()[name = tensor<string, []>("op_452"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<bool, []> qk_7_transpose_x_0 = const()[name = tensor<string, []>("qk_7_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> qk_7_transpose_y_0 = const()[name = tensor<string, []>("qk_7_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<int32, [4]> transpose_18_perm_0 = const()[name = tensor<string, []>("transpose_18_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [4]> transpose_19_perm_0 = const()[name = tensor<string, []>("transpose_19_perm_0"), val = tensor<int32, [4]>([0, 2, 3, 1])];
-            tensor<fp16, [1, 8, 64, 1500]> transpose_33 = transpose(perm = transpose_19_perm_0, x = k_15_cast);
-            tensor<fp16, [1, 8, 1500, 64]> transpose_34 = transpose(perm = transpose_18_perm_0, x = q_15_cast);
-            tensor<fp16, [1, 8, 1500, 1500]> qk_7_cast = matmul(transpose_x = qk_7_transpose_x_0, transpose_y = qk_7_transpose_y_0, x = transpose_34, y = transpose_33);
-            tensor<fp16, [1, 8, 1500, 1500]> var_456_cast = softmax(axis = var_391, x = qk_7_cast);
-            tensor<bool, []> var_458_transpose_x_0 = const()[name = tensor<string, []>("op_458_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> var_458_transpose_y_0 = const()[name = tensor<string, []>("op_458_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [1, 8, 1500, 64]> transpose_35 = transpose(perm = var_452, x = var_451_cast);
-            tensor<fp16, [1, 8, 1500, 64]> var_458_cast = matmul(transpose_x = var_458_transpose_x_0, transpose_y = var_458_transpose_y_0, x = var_456_cast, y = transpose_35);
-            tensor<int32, [4]> var_459 = const()[name = tensor<string, []>("op_459"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [3]> concat_3 = const()[name = tensor<string, []>("concat_3"), val = tensor<int32, [3]>([1, 1500, 512])];
-            tensor<fp16, [1, 1500, 8, 64]> transpose_32 = transpose(perm = var_459, x = var_458_cast);
-            tensor<fp16, [1, 1500, 512]> x_47_cast = reshape(shape = concat_3, x = transpose_32);
-            tensor<fp16, [512, 512]> var_464_to_fp16 = const()[name = tensor<string, []>("op_464_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(23852928)))];
-            tensor<fp16, [512]> var_465_to_fp16 = const()[name = tensor<string, []>("op_465_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24377280)))];
-            tensor<fp16, [1, 1500, 512]> var_466_cast = linear(bias = var_465_to_fp16, weight = var_464_to_fp16, x = x_47_cast);
-            tensor<fp16, [1, 1500, 512]> x_49_cast = add(x = x_43_cast, y = var_466_cast);
-            tensor<int32, [1]> var_472_axes_0 = const()[name = tensor<string, []>("op_472_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_3_mlp_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_3_mlp_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24378368)))];
-            tensor<fp16, [512]> blocks_3_mlp_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_3_mlp_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24379456)))];
-            tensor<fp16, [1, 1500, 512]> var_472_cast = layer_norm(axes = var_472_axes_0, beta = blocks_3_mlp_ln_bias_to_fp16, epsilon = var_397_to_fp16, gamma = blocks_3_mlp_ln_weight_to_fp16, x = x_49_cast);
-            tensor<fp16, [2048, 512]> var_481_to_fp16 = const()[name = tensor<string, []>("op_481_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24380544)))];
-            tensor<fp16, [2048]> var_482_to_fp16 = const()[name = tensor<string, []>("op_482_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(26477760)))];
-            tensor<fp16, [1, 1500, 2048]> input_33_cast = linear(bias = var_482_to_fp16, weight = var_481_to_fp16, x = var_472_cast);
-            tensor<string, []> x_53_mode_0 = const()[name = tensor<string, []>("x_53_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 1500, 2048]> x_53_cast = gelu(mode = x_53_mode_0, x = input_33_cast);
-            tensor<fp16, [512, 2048]> var_487_to_fp16 = const()[name = tensor<string, []>("op_487_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(26481920)))];
-            tensor<fp16, [512]> var_488_to_fp16 = const()[name = tensor<string, []>("op_488_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(28579136)))];
-            tensor<fp16, [1, 1500, 512]> var_489_cast = linear(bias = var_488_to_fp16, weight = var_487_to_fp16, x = x_53_cast);
-            tensor<fp16, [1, 1500, 512]> x_55_cast = add(x = x_49_cast, y = var_489_cast);
-            tensor<int32, []> var_498 = const()[name = tensor<string, []>("op_498"), val = tensor<int32, []>(-1)];
-            tensor<int32, [1]> var_515_axes_0 = const()[name = tensor<string, []>("op_515_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_4_attn_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_4_attn_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(28580224)))];
-            tensor<fp16, [512]> blocks_4_attn_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_4_attn_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(28581312)))];
-            tensor<fp16, []> var_504_to_fp16 = const()[name = tensor<string, []>("op_504_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 512]> var_515_cast = layer_norm(axes = var_515_axes_0, beta = blocks_4_attn_ln_bias_to_fp16, epsilon = var_504_to_fp16, gamma = blocks_4_attn_ln_weight_to_fp16, x = x_55_cast);
-            tensor<fp16, [512, 512]> var_526_to_fp16 = const()[name = tensor<string, []>("op_526_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(28582400)))];
-            tensor<fp16, [512]> var_527_to_fp16 = const()[name = tensor<string, []>("op_527_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(29106752)))];
-            tensor<fp16, [1, 1500, 512]> q_17_cast = linear(bias = var_527_to_fp16, weight = var_526_to_fp16, x = var_515_cast);
-            tensor<fp16, [512, 512]> var_530_to_fp16 = const()[name = tensor<string, []>("op_530_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(29107840)))];
-            tensor<fp16, [512]> k_17_bias_0_to_fp16 = const()[name = tensor<string, []>("k_17_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(29632192)))];
-            tensor<fp16, [1, 1500, 512]> k_17_cast = linear(bias = k_17_bias_0_to_fp16, weight = var_530_to_fp16, x = var_515_cast);
-            tensor<fp16, [512, 512]> var_534_to_fp16 = const()[name = tensor<string, []>("op_534_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(29633280)))];
-            tensor<fp16, [512]> var_535_to_fp16 = const()[name = tensor<string, []>("op_535_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(30157632)))];
-            tensor<fp16, [1, 1500, 512]> v_17_cast = linear(bias = var_535_to_fp16, weight = var_534_to_fp16, x = var_515_cast);
-            tensor<int32, [4]> var_543 = const()[name = tensor<string, []>("op_543"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_544_cast = reshape(shape = var_543, x = q_17_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_50_to_fp16 = const()[name = tensor<string, []>("const_50_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> q_19_cast = mul(x = var_544_cast, y = const_50_to_fp16);
-            tensor<int32, [4]> var_550 = const()[name = tensor<string, []>("op_550"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_551_cast = reshape(shape = var_550, x = k_17_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_51_to_fp16 = const()[name = tensor<string, []>("const_51_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> k_19_cast = mul(x = var_551_cast, y = const_51_to_fp16);
-            tensor<int32, [4]> var_557 = const()[name = tensor<string, []>("op_557"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_558_cast = reshape(shape = var_557, x = v_17_cast);
-            tensor<int32, [4]> var_559 = const()[name = tensor<string, []>("op_559"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<bool, []> qk_9_transpose_x_0 = const()[name = tensor<string, []>("qk_9_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> qk_9_transpose_y_0 = const()[name = tensor<string, []>("qk_9_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<int32, [4]> transpose_20_perm_0 = const()[name = tensor<string, []>("transpose_20_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [4]> transpose_21_perm_0 = const()[name = tensor<string, []>("transpose_21_perm_0"), val = tensor<int32, [4]>([0, 2, 3, 1])];
-            tensor<fp16, [1, 8, 64, 1500]> transpose_29 = transpose(perm = transpose_21_perm_0, x = k_19_cast);
-            tensor<fp16, [1, 8, 1500, 64]> transpose_30 = transpose(perm = transpose_20_perm_0, x = q_19_cast);
-            tensor<fp16, [1, 8, 1500, 1500]> qk_9_cast = matmul(transpose_x = qk_9_transpose_x_0, transpose_y = qk_9_transpose_y_0, x = transpose_30, y = transpose_29);
-            tensor<fp16, [1, 8, 1500, 1500]> var_563_cast = softmax(axis = var_498, x = qk_9_cast);
-            tensor<bool, []> var_565_transpose_x_0 = const()[name = tensor<string, []>("op_565_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> var_565_transpose_y_0 = const()[name = tensor<string, []>("op_565_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [1, 8, 1500, 64]> transpose_31 = transpose(perm = var_559, x = var_558_cast);
-            tensor<fp16, [1, 8, 1500, 64]> var_565_cast = matmul(transpose_x = var_565_transpose_x_0, transpose_y = var_565_transpose_y_0, x = var_563_cast, y = transpose_31);
-            tensor<int32, [4]> var_566 = const()[name = tensor<string, []>("op_566"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [3]> concat_4 = const()[name = tensor<string, []>("concat_4"), val = tensor<int32, [3]>([1, 1500, 512])];
-            tensor<fp16, [1, 1500, 8, 64]> transpose_28 = transpose(perm = var_566, x = var_565_cast);
-            tensor<fp16, [1, 1500, 512]> x_59_cast = reshape(shape = concat_4, x = transpose_28);
-            tensor<fp16, [512, 512]> var_571_to_fp16 = const()[name = tensor<string, []>("op_571_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(30158720)))];
-            tensor<fp16, [512]> var_572_to_fp16 = const()[name = tensor<string, []>("op_572_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(30683072)))];
-            tensor<fp16, [1, 1500, 512]> var_573_cast = linear(bias = var_572_to_fp16, weight = var_571_to_fp16, x = x_59_cast);
-            tensor<fp16, [1, 1500, 512]> x_61_cast = add(x = x_55_cast, y = var_573_cast);
-            tensor<int32, [1]> var_579_axes_0 = const()[name = tensor<string, []>("op_579_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_4_mlp_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_4_mlp_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(30684160)))];
-            tensor<fp16, [512]> blocks_4_mlp_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_4_mlp_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(30685248)))];
-            tensor<fp16, [1, 1500, 512]> var_579_cast = layer_norm(axes = var_579_axes_0, beta = blocks_4_mlp_ln_bias_to_fp16, epsilon = var_504_to_fp16, gamma = blocks_4_mlp_ln_weight_to_fp16, x = x_61_cast);
-            tensor<fp16, [2048, 512]> var_588_to_fp16 = const()[name = tensor<string, []>("op_588_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(30686336)))];
-            tensor<fp16, [2048]> var_589_to_fp16 = const()[name = tensor<string, []>("op_589_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(32783552)))];
-            tensor<fp16, [1, 1500, 2048]> input_41_cast = linear(bias = var_589_to_fp16, weight = var_588_to_fp16, x = var_579_cast);
-            tensor<string, []> x_65_mode_0 = const()[name = tensor<string, []>("x_65_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 1500, 2048]> x_65_cast = gelu(mode = x_65_mode_0, x = input_41_cast);
-            tensor<fp16, [512, 2048]> var_594_to_fp16 = const()[name = tensor<string, []>("op_594_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(32787712)))];
-            tensor<fp16, [512]> var_595_to_fp16 = const()[name = tensor<string, []>("op_595_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34884928)))];
-            tensor<fp16, [1, 1500, 512]> var_596_cast = linear(bias = var_595_to_fp16, weight = var_594_to_fp16, x = x_65_cast);
-            tensor<fp16, [1, 1500, 512]> x_67_cast = add(x = x_61_cast, y = var_596_cast);
-            tensor<int32, []> var_605 = const()[name = tensor<string, []>("op_605"), val = tensor<int32, []>(-1)];
-            tensor<int32, [1]> var_622_axes_0 = const()[name = tensor<string, []>("op_622_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_5_attn_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_5_attn_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34886016)))];
-            tensor<fp16, [512]> blocks_5_attn_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_5_attn_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34887104)))];
-            tensor<fp16, []> var_611_to_fp16 = const()[name = tensor<string, []>("op_611_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 512]> var_622_cast = layer_norm(axes = var_622_axes_0, beta = blocks_5_attn_ln_bias_to_fp16, epsilon = var_611_to_fp16, gamma = blocks_5_attn_ln_weight_to_fp16, x = x_67_cast);
-            tensor<fp16, [512, 512]> var_633_to_fp16 = const()[name = tensor<string, []>("op_633_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34888192)))];
-            tensor<fp16, [512]> var_634_to_fp16 = const()[name = tensor<string, []>("op_634_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(35412544)))];
-            tensor<fp16, [1, 1500, 512]> q_21_cast = linear(bias = var_634_to_fp16, weight = var_633_to_fp16, x = var_622_cast);
-            tensor<fp16, [512, 512]> var_637_to_fp16 = const()[name = tensor<string, []>("op_637_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(35413632)))];
-            tensor<fp16, [512]> k_21_bias_0_to_fp16 = const()[name = tensor<string, []>("k_21_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(35937984)))];
-            tensor<fp16, [1, 1500, 512]> k_21_cast = linear(bias = k_21_bias_0_to_fp16, weight = var_637_to_fp16, x = var_622_cast);
-            tensor<fp16, [512, 512]> var_641_to_fp16 = const()[name = tensor<string, []>("op_641_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(35939072)))];
-            tensor<fp16, [512]> var_642_to_fp16 = const()[name = tensor<string, []>("op_642_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36463424)))];
-            tensor<fp16, [1, 1500, 512]> v_21_cast = linear(bias = var_642_to_fp16, weight = var_641_to_fp16, x = var_622_cast);
-            tensor<int32, [4]> var_650 = const()[name = tensor<string, []>("op_650"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_651_cast = reshape(shape = var_650, x = q_21_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_52_to_fp16 = const()[name = tensor<string, []>("const_52_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> q_cast = mul(x = var_651_cast, y = const_52_to_fp16);
-            tensor<int32, [4]> var_657 = const()[name = tensor<string, []>("op_657"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_658_cast = reshape(shape = var_657, x = k_21_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_53_to_fp16 = const()[name = tensor<string, []>("const_53_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 8, 64]> k_cast = mul(x = var_658_cast, y = const_53_to_fp16);
-            tensor<int32, [4]> var_664 = const()[name = tensor<string, []>("op_664"), val = tensor<int32, [4]>([1, 1500, 8, -1])];
-            tensor<fp16, [1, 1500, 8, 64]> var_665_cast = reshape(shape = var_664, x = v_21_cast);
-            tensor<int32, [4]> var_666 = const()[name = tensor<string, []>("op_666"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<bool, []> qk_transpose_x_0 = const()[name = tensor<string, []>("qk_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> qk_transpose_y_0 = const()[name = tensor<string, []>("qk_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<int32, [4]> transpose_22_perm_0 = const()[name = tensor<string, []>("transpose_22_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [4]> transpose_23_perm_0 = const()[name = tensor<string, []>("transpose_23_perm_0"), val = tensor<int32, [4]>([0, 2, 3, 1])];
-            tensor<fp16, [1, 8, 64, 1500]> transpose_25 = transpose(perm = transpose_23_perm_0, x = k_cast);
-            tensor<fp16, [1, 8, 1500, 64]> transpose_26 = transpose(perm = transpose_22_perm_0, x = q_cast);
-            tensor<fp16, [1, 8, 1500, 1500]> qk_cast = matmul(transpose_x = qk_transpose_x_0, transpose_y = qk_transpose_y_0, x = transpose_26, y = transpose_25);
-            tensor<fp16, [1, 8, 1500, 1500]> var_670_cast = softmax(axis = var_605, x = qk_cast);
-            tensor<bool, []> var_672_transpose_x_0 = const()[name = tensor<string, []>("op_672_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> var_672_transpose_y_0 = const()[name = tensor<string, []>("op_672_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [1, 8, 1500, 64]> transpose_27 = transpose(perm = var_666, x = var_665_cast);
-            tensor<fp16, [1, 8, 1500, 64]> var_672_cast = matmul(transpose_x = var_672_transpose_x_0, transpose_y = var_672_transpose_y_0, x = var_670_cast, y = transpose_27);
-            tensor<int32, [4]> var_673 = const()[name = tensor<string, []>("op_673"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [3]> concat_5 = const()[name = tensor<string, []>("concat_5"), val = tensor<int32, [3]>([1, 1500, 512])];
-            tensor<fp16, [1, 1500, 8, 64]> transpose_24 = transpose(perm = var_673, x = var_672_cast);
-            tensor<fp16, [1, 1500, 512]> x_71_cast = reshape(shape = concat_5, x = transpose_24);
-            tensor<fp16, [512, 512]> var_678_to_fp16 = const()[name = tensor<string, []>("op_678_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36464512)))];
-            tensor<fp16, [512]> var_679_to_fp16 = const()[name = tensor<string, []>("op_679_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36988864)))];
-            tensor<fp16, [1, 1500, 512]> var_680_cast = linear(bias = var_679_to_fp16, weight = var_678_to_fp16, x = x_71_cast);
-            tensor<fp16, [1, 1500, 512]> x_73_cast = add(x = x_67_cast, y = var_680_cast);
-            tensor<int32, [1]> var_686_axes_0 = const()[name = tensor<string, []>("op_686_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> blocks_5_mlp_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_5_mlp_ln_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36989952)))];
-            tensor<fp16, [512]> blocks_5_mlp_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_5_mlp_ln_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36991040)))];
-            tensor<fp16, [1, 1500, 512]> var_686_cast = layer_norm(axes = var_686_axes_0, beta = blocks_5_mlp_ln_bias_to_fp16, epsilon = var_611_to_fp16, gamma = blocks_5_mlp_ln_weight_to_fp16, x = x_73_cast);
-            tensor<fp16, [2048, 512]> var_695_to_fp16 = const()[name = tensor<string, []>("op_695_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36992128)))];
-            tensor<fp16, [2048]> var_696_to_fp16 = const()[name = tensor<string, []>("op_696_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(39089344)))];
-            tensor<fp16, [1, 1500, 2048]> input_49_cast = linear(bias = var_696_to_fp16, weight = var_695_to_fp16, x = var_686_cast);
-            tensor<string, []> x_77_mode_0 = const()[name = tensor<string, []>("x_77_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 1500, 2048]> x_77_cast = gelu(mode = x_77_mode_0, x = input_49_cast);
-            tensor<fp16, [512, 2048]> var_701_to_fp16 = const()[name = tensor<string, []>("op_701_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(39093504)))];
-            tensor<fp16, [512]> var_702_to_fp16 = const()[name = tensor<string, []>("op_702_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(41190720)))];
-            tensor<fp16, [1, 1500, 512]> var_703_cast = linear(bias = var_702_to_fp16, weight = var_701_to_fp16, x = x_77_cast);
-            tensor<fp16, [1, 1500, 512]> x_cast = add(x = x_73_cast, y = var_703_cast);
-            tensor<int32, [1]> var_716_axes_0 = const()[name = tensor<string, []>("op_716_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [512]> ln_post_weight_to_fp16 = const()[name = tensor<string, []>("ln_post_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(41191808)))];
-            tensor<fp16, [512]> ln_post_bias_to_fp16 = const()[name = tensor<string, []>("ln_post_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(41192896)))];
-            tensor<fp16, []> var_707_to_fp16 = const()[name = tensor<string, []>("op_707_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 512]> var_716_cast = layer_norm(axes = var_716_axes_0, beta = ln_post_bias_to_fp16, epsilon = var_707_to_fp16, gamma = ln_post_weight_to_fp16, x = x_cast);
-            tensor<string, []> var_716_cast_to_fp32_dtype_0 = const()[name = tensor<string, []>("op_716_cast_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
-            tensor<fp32, [1, 1500, 512]> output = cast(dtype = var_716_cast_to_fp32_dtype_0, x = var_716_cast);
-        } -> (output);
-}

ggml-large-encoder.mlmodelc.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e52af78363b58219ac2add6df9bddd49766292ef7767a5f6a9b6617052ec52d9
+size 1174643319

ggml-large-encoder.mlmodelc/coremldata.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:05fe28591b40616fa0c34ad7b853133623f5300923ec812acb11459c411acf3b
-size 149

ggml-large-encoder.mlmodelc/metadata.json

@@ -1,64 +0,0 @@
-[
-  {
-    "metadataOutputVersion" : "3.0",
-    "storagePrecision" : "Float16",
-    "outputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32)",
-        "shortDescription" : "",
-        "shape" : "[]",
-        "name" : "output",
-        "type" : "MultiArray"
-      }
-    ],
-    "modelParameters" : [
-
-    ],
-    "specificationVersion" : 6,
-    "mlProgramOperationTypeHistogram" : {
-      "Linear" : 192,
-      "Matmul" : 64,
-      "Cast" : 2,
-      "Conv" : 2,
-      "Softmax" : 32,
-      "Add" : 65,
-      "LayerNorm" : 65,
-      "Mul" : 64,
-      "Transpose" : 129,
-      "Gelu" : 34,
-      "Reshape" : 128
-    },
-    "computePrecision" : "Mixed (Float16, Float32, Int32)",
-    "isUpdatable" : "0",
-    "availability" : {
-      "macOS" : "12.0",
-      "tvOS" : "15.0",
-      "watchOS" : "8.0",
-      "iOS" : "15.0",
-      "macCatalyst" : "15.0"
-    },
-    "modelType" : {
-      "name" : "MLModelType_mlProgram"
-    },
-    "userDefinedMetadata" : {
-
-    },
-    "inputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32 1 × 80 × 3000)",
-        "shortDescription" : "",
-        "shape" : "[1, 80, 3000]",
-        "name" : "logmel_data",
-        "type" : "MultiArray"
-      }
-    ],
-    "generatedClassName" : "coreml_encoder_large",
-    "method" : "predict"
-  }
-]

ggml-large-encoder.mlmodelc/weights/weight.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:774d50f0f4191a3bb23c3bffe24367dac5a6695cf3f099ef1acf10c992d96fa0
-size 1273684480

ggml-base-encoder.mlmodelc/weights/weight.bin → ggml-medium-encoder.mlmodelc.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f7be921371a600ad5fa7e706773bdf0cf7b833f95ea384cc5aa1afaa31151acd
-size 41193984
+oid sha256:dbe6b96b61874e0d9536ae31eb13ae294c820d8e8cd4aa700fa916e93f926099
+size 567829197

ggml-medium-encoder.mlmodelc/analytics/coremldata.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:adbe456375e7eb3407732a426ecb65bbda86860e4aa801f3a696b70b8a533cdd
-size 207

ggml-medium-encoder.mlmodelc/coremldata.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:05fe28591b40616fa0c34ad7b853133623f5300923ec812acb11459c411acf3b
-size 149

ggml-medium-encoder.mlmodelc/metadata.json

@@ -1,64 +0,0 @@
-[
-  {
-    "metadataOutputVersion" : "3.0",
-    "storagePrecision" : "Float16",
-    "outputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32)",
-        "shortDescription" : "",
-        "shape" : "[]",
-        "name" : "output",
-        "type" : "MultiArray"
-      }
-    ],
-    "modelParameters" : [
-
-    ],
-    "specificationVersion" : 6,
-    "mlProgramOperationTypeHistogram" : {
-      "Linear" : 144,
-      "Matmul" : 48,
-      "Cast" : 2,
-      "Conv" : 2,
-      "Softmax" : 24,
-      "Add" : 49,
-      "LayerNorm" : 49,
-      "Mul" : 48,
-      "Transpose" : 97,
-      "Gelu" : 26,
-      "Reshape" : 96
-    },
-    "computePrecision" : "Mixed (Float16, Float32, Int32)",
-    "isUpdatable" : "0",
-    "availability" : {
-      "macOS" : "12.0",
-      "tvOS" : "15.0",
-      "watchOS" : "8.0",
-      "iOS" : "15.0",
-      "macCatalyst" : "15.0"
-    },
-    "modelType" : {
-      "name" : "MLModelType_mlProgram"
-    },
-    "userDefinedMetadata" : {
-
-    },
-    "inputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32 1 × 80 × 3000)",
-        "shortDescription" : "",
-        "shape" : "[1, 80, 3000]",
-        "name" : "logmel_data",
-        "type" : "MultiArray"
-      }
-    ],
-    "generatedClassName" : "coreml_encoder_medium",
-    "method" : "predict"
-  }
-]

ggml-medium-encoder.mlmodelc/weights/weight.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:111548c34d144aa0647caae75769976723977dc6f4d60ef94c2f3e82d9ecc0e2
-size 614507008

ggml-large-encoder.mlmodelc/analytics/coremldata.bin → ggml-medium.en-encoder.mlmodelc.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5aec358d66255e1eecb43c5df3acc31815bed1b5d23ec4caa89cb7dfe1b89cc4
-size 207
+oid sha256:5efbabd0c02344087a7bd0ffa42116507746a1e6023ff9665b7f58dabe23b6d4
+size 566993104

ggml-medium.en-encoder.mlmodelc/analytics/coremldata.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:adbe456375e7eb3407732a426ecb65bbda86860e4aa801f3a696b70b8a533cdd
-size 207

ggml-medium.en-encoder.mlmodelc/coremldata.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:05fe28591b40616fa0c34ad7b853133623f5300923ec812acb11459c411acf3b
-size 149

ggml-medium.en-encoder.mlmodelc/metadata.json

@@ -1,64 +0,0 @@
-[
-  {
-    "metadataOutputVersion" : "3.0",
-    "storagePrecision" : "Float16",
-    "outputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32)",
-        "shortDescription" : "",
-        "shape" : "[]",
-        "name" : "output",
-        "type" : "MultiArray"
-      }
-    ],
-    "modelParameters" : [
-
-    ],
-    "specificationVersion" : 6,
-    "mlProgramOperationTypeHistogram" : {
-      "Linear" : 144,
-      "Matmul" : 48,
-      "Cast" : 2,
-      "Conv" : 2,
-      "Softmax" : 24,
-      "Add" : 49,
-      "LayerNorm" : 49,
-      "Mul" : 48,
-      "Transpose" : 97,
-      "Gelu" : 26,
-      "Reshape" : 96
-    },
-    "computePrecision" : "Mixed (Float16, Float32, Int32)",
-    "isUpdatable" : "0",
-    "availability" : {
-      "macOS" : "12.0",
-      "tvOS" : "15.0",
-      "watchOS" : "8.0",
-      "iOS" : "15.0",
-      "macCatalyst" : "15.0"
-    },
-    "modelType" : {
-      "name" : "MLModelType_mlProgram"
-    },
-    "userDefinedMetadata" : {
-
-    },
-    "inputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32 1 × 80 × 3000)",
-        "shortDescription" : "",
-        "shape" : "[1, 80, 3000]",
-        "name" : "logmel_data",
-        "type" : "MultiArray"
-      }
-    ],
-    "generatedClassName" : "coreml_encoder_medium_en",
-    "method" : "predict"
-  }
-]

ggml-medium.en-encoder.mlmodelc/weights/weight.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:dbabff0d30bb8565baf63618dc2c2eb6fb7e8ef2fe3d4114b68c26e784b91398
-size 614507008

ggml-small-encoder.mlmodelc.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5305b8521b3766bbfb9a042f92f7986f5d005271247ed6e872f1634dc90d22d5
+size 163083036

ggml-small-encoder.mlmodelc/analytics/coremldata.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:18ad2072ae82872c2ba8a187071e1e7d6c1105253685e7aa95138adcf07874e0
-size 207

ggml-small-encoder.mlmodelc/coremldata.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:05fe28591b40616fa0c34ad7b853133623f5300923ec812acb11459c411acf3b
-size 149

ggml-small-encoder.mlmodelc/metadata.json

@@ -1,64 +0,0 @@
-[
-  {
-    "metadataOutputVersion" : "3.0",
-    "storagePrecision" : "Float16",
-    "outputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32)",
-        "shortDescription" : "",
-        "shape" : "[]",
-        "name" : "output",
-        "type" : "MultiArray"
-      }
-    ],
-    "modelParameters" : [
-
-    ],
-    "specificationVersion" : 6,
-    "mlProgramOperationTypeHistogram" : {
-      "Linear" : 72,
-      "Matmul" : 24,
-      "Cast" : 2,
-      "Conv" : 2,
-      "Softmax" : 12,
-      "Add" : 25,
-      "LayerNorm" : 25,
-      "Mul" : 24,
-      "Transpose" : 49,
-      "Gelu" : 14,
-      "Reshape" : 48
-    },
-    "computePrecision" : "Mixed (Float16, Float32, Int32)",
-    "isUpdatable" : "0",
-    "availability" : {
-      "macOS" : "12.0",
-      "tvOS" : "15.0",
-      "watchOS" : "8.0",
-      "iOS" : "15.0",
-      "macCatalyst" : "15.0"
-    },
-    "modelType" : {
-      "name" : "MLModelType_mlProgram"
-    },
-    "userDefinedMetadata" : {
-
-    },
-    "inputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32 1 × 80 × 3000)",
-        "shortDescription" : "",
-        "shape" : "[1, 80, 3000]",
-        "name" : "logmel_data",
-        "type" : "MultiArray"
-      }
-    ],
-    "generatedClassName" : "coreml_encoder_small",
-    "method" : "predict"
-  }
-]

ggml-small-encoder.mlmodelc/weights/weight.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:517bd4dc214ae6ab9db4cab7b29873d4d5a1453e513e0aa8634402ddc345e6f8
-size 176339456

ggml-base-encoder.mlmodelc/coremldata.bin → ggml-tiny-encoder.mlmodelc.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:05fe28591b40616fa0c34ad7b853133623f5300923ec812acb11459c411acf3b
-size 149
+oid sha256:3c21c0f5584b80ca94d7a6fac8d76a93c9ceb38750c3275775cad3161132f03b
+size 15037455

ggml-tiny-encoder.mlmodelc/analytics/coremldata.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:f2f746e204d04779629b4959dad86d4ada040edd54f0ed1c80aa6f482a0fa72e
-size 207

ggml-tiny-encoder.mlmodelc/coremldata.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:05fe28591b40616fa0c34ad7b853133623f5300923ec812acb11459c411acf3b
-size 149

ggml-tiny-encoder.mlmodelc/metadata.json

@@ -1,64 +0,0 @@
-[
-  {
-    "metadataOutputVersion" : "3.0",
-    "storagePrecision" : "Float16",
-    "outputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32)",
-        "shortDescription" : "",
-        "shape" : "[]",
-        "name" : "output",
-        "type" : "MultiArray"
-      }
-    ],
-    "modelParameters" : [
-
-    ],
-    "specificationVersion" : 6,
-    "mlProgramOperationTypeHistogram" : {
-      "Linear" : 24,
-      "Matmul" : 8,
-      "Cast" : 2,
-      "Conv" : 2,
-      "Softmax" : 4,
-      "Add" : 9,
-      "LayerNorm" : 9,
-      "Mul" : 8,
-      "Transpose" : 17,
-      "Gelu" : 6,
-      "Reshape" : 16
-    },
-    "computePrecision" : "Mixed (Float16, Float32, Int32)",
-    "isUpdatable" : "0",
-    "availability" : {
-      "macOS" : "12.0",
-      "tvOS" : "15.0",
-      "watchOS" : "8.0",
-      "iOS" : "15.0",
-      "macCatalyst" : "15.0"
-    },
-    "modelType" : {
-      "name" : "MLModelType_mlProgram"
-    },
-    "userDefinedMetadata" : {
-
-    },
-    "inputSchema" : [
-      {
-        "hasShapeFlexibility" : "0",
-        "isOptional" : "0",
-        "dataType" : "Float32",
-        "formattedType" : "MultiArray (Float32 1 × 80 × 3000)",
-        "shortDescription" : "",
-        "shape" : "[1, 80, 3000]",
-        "name" : "logmel_data",
-        "type" : "MultiArray"
-      }
-    ],
-    "generatedClassName" : "coreml_encoder_tiny",
-    "method" : "predict"
-  }
-]

ggml-tiny-encoder.mlmodelc/model.mil

@@ -1,275 +0,0 @@
-program(1.0)
-[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "4.28.4"}, {"coremlc-version", "1436.100.10"}})]
-{
-    func main<ios15>(tensor<fp32, [1, 80, 3000]> logmel_data) {
-            tensor<int32, []> var_16 = const()[name = tensor<string, []>("op_16"), val = tensor<int32, []>(1)];
-            tensor<int32, [1]> var_24 = const()[name = tensor<string, []>("op_24"), val = tensor<int32, [1]>([1])];
-            tensor<int32, [1]> var_26 = const()[name = tensor<string, []>("op_26"), val = tensor<int32, [1]>([1])];
-            tensor<string, []> var_28_pad_type_0 = const()[name = tensor<string, []>("op_28_pad_type_0"), val = tensor<string, []>("custom")];
-            tensor<int32, [2]> var_28_pad_0 = const()[name = tensor<string, []>("op_28_pad_0"), val = tensor<int32, [2]>([1, 1])];
-            tensor<string, []> logmel_data_to_fp16_dtype_0 = const()[name = tensor<string, []>("logmel_data_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
-            tensor<fp16, [384, 80, 3]> weight_3_to_fp16 = const()[name = tensor<string, []>("weight_3_to_fp16"), val = tensor<fp16, [384, 80, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
-            tensor<fp16, [384]> bias_3_to_fp16 = const()[name = tensor<string, []>("bias_3_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(184448)))];
-            tensor<fp16, [1, 80, 3000]> cast_127 = cast(dtype = logmel_data_to_fp16_dtype_0, x = logmel_data);
-            tensor<fp16, [1, 384, 3000]> var_28_cast = conv(bias = bias_3_to_fp16, dilations = var_26, groups = var_16, pad = var_28_pad_0, pad_type = var_28_pad_type_0, strides = var_24, weight = weight_3_to_fp16, x = cast_127);
-            tensor<string, []> input_1_mode_0 = const()[name = tensor<string, []>("input_1_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 384, 3000]> input_1_cast = gelu(mode = input_1_mode_0, x = var_28_cast);
-            tensor<int32, []> var_32 = const()[name = tensor<string, []>("op_32"), val = tensor<int32, []>(1)];
-            tensor<int32, [1]> var_41 = const()[name = tensor<string, []>("op_41"), val = tensor<int32, [1]>([2])];
-            tensor<int32, [1]> var_43 = const()[name = tensor<string, []>("op_43"), val = tensor<int32, [1]>([1])];
-            tensor<string, []> var_45_pad_type_0 = const()[name = tensor<string, []>("op_45_pad_type_0"), val = tensor<string, []>("custom")];
-            tensor<int32, [2]> var_45_pad_0 = const()[name = tensor<string, []>("op_45_pad_0"), val = tensor<int32, [2]>([1, 1])];
-            tensor<fp16, [384, 384, 3]> weight_7_to_fp16 = const()[name = tensor<string, []>("weight_7_to_fp16"), val = tensor<fp16, [384, 384, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(185280)))];
-            tensor<fp16, [384]> bias_7_to_fp16 = const()[name = tensor<string, []>("bias_7_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1070080)))];
-            tensor<fp16, [1, 384, 1500]> var_45_cast = conv(bias = bias_7_to_fp16, dilations = var_43, groups = var_32, pad = var_45_pad_0, pad_type = var_45_pad_type_0, strides = var_41, weight = weight_7_to_fp16, x = input_1_cast);
-            tensor<string, []> x_3_mode_0 = const()[name = tensor<string, []>("x_3_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 384, 1500]> x_3_cast = gelu(mode = x_3_mode_0, x = var_45_cast);
-            tensor<int32, [3]> var_50 = const()[name = tensor<string, []>("op_50"), val = tensor<int32, [3]>([0, 2, 1])];
-            tensor<fp16, [1500, 384]> positional_embedding_to_fp16 = const()[name = tensor<string, []>("positional_embedding_to_fp16"), val = tensor<fp16, [1500, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1070912)))];
-            tensor<fp16, [1, 1500, 384]> transpose_32 = transpose(perm = var_50, x = x_3_cast);
-            tensor<fp16, [1, 1500, 384]> var_53_cast = add(x = transpose_32, y = positional_embedding_to_fp16);
-            tensor<int32, []> var_65 = const()[name = tensor<string, []>("op_65"), val = tensor<int32, []>(-1)];
-            tensor<int32, [1]> var_82_axes_0 = const()[name = tensor<string, []>("op_82_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [384]> blocks_0_attn_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_0_attn_ln_weight_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2222976)))];
-            tensor<fp16, [384]> blocks_0_attn_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_0_attn_ln_bias_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2223808)))];
-            tensor<fp16, []> var_71_to_fp16 = const()[name = tensor<string, []>("op_71_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 384]> var_82_cast = layer_norm(axes = var_82_axes_0, beta = blocks_0_attn_ln_bias_to_fp16, epsilon = var_71_to_fp16, gamma = blocks_0_attn_ln_weight_to_fp16, x = var_53_cast);
-            tensor<fp16, [384, 384]> var_93_to_fp16 = const()[name = tensor<string, []>("op_93_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2224640)))];
-            tensor<fp16, [384]> var_94_to_fp16 = const()[name = tensor<string, []>("op_94_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2519616)))];
-            tensor<fp16, [1, 1500, 384]> q_1_cast = linear(bias = var_94_to_fp16, weight = var_93_to_fp16, x = var_82_cast);
-            tensor<fp16, [384, 384]> var_97_to_fp16 = const()[name = tensor<string, []>("op_97_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2520448)))];
-            tensor<fp16, [384]> k_1_bias_0_to_fp16 = const()[name = tensor<string, []>("k_1_bias_0_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2815424)))];
-            tensor<fp16, [1, 1500, 384]> k_1_cast = linear(bias = k_1_bias_0_to_fp16, weight = var_97_to_fp16, x = var_82_cast);
-            tensor<fp16, [384, 384]> var_101_to_fp16 = const()[name = tensor<string, []>("op_101_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2816256)))];
-            tensor<fp16, [384]> var_102_to_fp16 = const()[name = tensor<string, []>("op_102_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3111232)))];
-            tensor<fp16, [1, 1500, 384]> v_1_cast = linear(bias = var_102_to_fp16, weight = var_101_to_fp16, x = var_82_cast);
-            tensor<int32, [4]> var_110 = const()[name = tensor<string, []>("op_110"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_111_cast = reshape(shape = var_110, x = q_1_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_28_to_fp16 = const()[name = tensor<string, []>("const_28_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 6, 64]> q_3_cast = mul(x = var_111_cast, y = const_28_to_fp16);
-            tensor<int32, [4]> var_117 = const()[name = tensor<string, []>("op_117"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_118_cast = reshape(shape = var_117, x = k_1_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_29_to_fp16 = const()[name = tensor<string, []>("const_29_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 6, 64]> k_3_cast = mul(x = var_118_cast, y = const_29_to_fp16);
-            tensor<int32, [4]> var_124 = const()[name = tensor<string, []>("op_124"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_125_cast = reshape(shape = var_124, x = v_1_cast);
-            tensor<int32, [4]> var_126 = const()[name = tensor<string, []>("op_126"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<bool, []> qk_1_transpose_x_0 = const()[name = tensor<string, []>("qk_1_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> qk_1_transpose_y_0 = const()[name = tensor<string, []>("qk_1_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<int32, [4]> transpose_8_perm_0 = const()[name = tensor<string, []>("transpose_8_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [4]> transpose_9_perm_0 = const()[name = tensor<string, []>("transpose_9_perm_0"), val = tensor<int32, [4]>([0, 2, 3, 1])];
-            tensor<fp16, [1, 6, 64, 1500]> transpose_29 = transpose(perm = transpose_9_perm_0, x = k_3_cast);
-            tensor<fp16, [1, 6, 1500, 64]> transpose_30 = transpose(perm = transpose_8_perm_0, x = q_3_cast);
-            tensor<fp16, [1, 6, 1500, 1500]> qk_1_cast = matmul(transpose_x = qk_1_transpose_x_0, transpose_y = qk_1_transpose_y_0, x = transpose_30, y = transpose_29);
-            tensor<fp16, [1, 6, 1500, 1500]> var_130_cast = softmax(axis = var_65, x = qk_1_cast);
-            tensor<bool, []> var_132_transpose_x_0 = const()[name = tensor<string, []>("op_132_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> var_132_transpose_y_0 = const()[name = tensor<string, []>("op_132_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [1, 6, 1500, 64]> transpose_31 = transpose(perm = var_126, x = var_125_cast);
-            tensor<fp16, [1, 6, 1500, 64]> var_132_cast = matmul(transpose_x = var_132_transpose_x_0, transpose_y = var_132_transpose_y_0, x = var_130_cast, y = transpose_31);
-            tensor<int32, [4]> var_133 = const()[name = tensor<string, []>("op_133"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [3]> concat_0 = const()[name = tensor<string, []>("concat_0"), val = tensor<int32, [3]>([1, 1500, 384])];
-            tensor<fp16, [1, 1500, 6, 64]> transpose_28 = transpose(perm = var_133, x = var_132_cast);
-            tensor<fp16, [1, 1500, 384]> x_11_cast = reshape(shape = concat_0, x = transpose_28);
-            tensor<fp16, [384, 384]> var_138_to_fp16 = const()[name = tensor<string, []>("op_138_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3112064)))];
-            tensor<fp16, [384]> var_139_to_fp16 = const()[name = tensor<string, []>("op_139_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3407040)))];
-            tensor<fp16, [1, 1500, 384]> var_140_cast = linear(bias = var_139_to_fp16, weight = var_138_to_fp16, x = x_11_cast);
-            tensor<fp16, [1, 1500, 384]> x_13_cast = add(x = var_53_cast, y = var_140_cast);
-            tensor<int32, [1]> var_146_axes_0 = const()[name = tensor<string, []>("op_146_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [384]> blocks_0_mlp_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_0_mlp_ln_weight_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3407872)))];
-            tensor<fp16, [384]> blocks_0_mlp_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_0_mlp_ln_bias_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3408704)))];
-            tensor<fp16, [1, 1500, 384]> var_146_cast = layer_norm(axes = var_146_axes_0, beta = blocks_0_mlp_ln_bias_to_fp16, epsilon = var_71_to_fp16, gamma = blocks_0_mlp_ln_weight_to_fp16, x = x_13_cast);
-            tensor<fp16, [1536, 384]> var_155_to_fp16 = const()[name = tensor<string, []>("op_155_to_fp16"), val = tensor<fp16, [1536, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3409536)))];
-            tensor<fp16, [1536]> var_156_to_fp16 = const()[name = tensor<string, []>("op_156_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4589248)))];
-            tensor<fp16, [1, 1500, 1536]> input_9_cast = linear(bias = var_156_to_fp16, weight = var_155_to_fp16, x = var_146_cast);
-            tensor<string, []> x_17_mode_0 = const()[name = tensor<string, []>("x_17_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 1500, 1536]> x_17_cast = gelu(mode = x_17_mode_0, x = input_9_cast);
-            tensor<fp16, [384, 1536]> var_161_to_fp16 = const()[name = tensor<string, []>("op_161_to_fp16"), val = tensor<fp16, [384, 1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4592384)))];
-            tensor<fp16, [384]> var_162_to_fp16 = const()[name = tensor<string, []>("op_162_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5772096)))];
-            tensor<fp16, [1, 1500, 384]> var_163_cast = linear(bias = var_162_to_fp16, weight = var_161_to_fp16, x = x_17_cast);
-            tensor<fp16, [1, 1500, 384]> x_19_cast = add(x = x_13_cast, y = var_163_cast);
-            tensor<int32, []> var_171 = const()[name = tensor<string, []>("op_171"), val = tensor<int32, []>(-1)];
-            tensor<int32, [1]> var_188_axes_0 = const()[name = tensor<string, []>("op_188_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [384]> blocks_1_attn_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_1_attn_ln_weight_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5772928)))];
-            tensor<fp16, [384]> blocks_1_attn_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_1_attn_ln_bias_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5773760)))];
-            tensor<fp16, []> var_177_to_fp16 = const()[name = tensor<string, []>("op_177_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 384]> var_188_cast = layer_norm(axes = var_188_axes_0, beta = blocks_1_attn_ln_bias_to_fp16, epsilon = var_177_to_fp16, gamma = blocks_1_attn_ln_weight_to_fp16, x = x_19_cast);
-            tensor<fp16, [384, 384]> var_199_to_fp16 = const()[name = tensor<string, []>("op_199_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5774592)))];
-            tensor<fp16, [384]> var_200_to_fp16 = const()[name = tensor<string, []>("op_200_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6069568)))];
-            tensor<fp16, [1, 1500, 384]> q_5_cast = linear(bias = var_200_to_fp16, weight = var_199_to_fp16, x = var_188_cast);
-            tensor<fp16, [384, 384]> var_203_to_fp16 = const()[name = tensor<string, []>("op_203_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6070400)))];
-            tensor<fp16, [384]> k_5_bias_0_to_fp16 = const()[name = tensor<string, []>("k_5_bias_0_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6365376)))];
-            tensor<fp16, [1, 1500, 384]> k_5_cast = linear(bias = k_5_bias_0_to_fp16, weight = var_203_to_fp16, x = var_188_cast);
-            tensor<fp16, [384, 384]> var_207_to_fp16 = const()[name = tensor<string, []>("op_207_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6366208)))];
-            tensor<fp16, [384]> var_208_to_fp16 = const()[name = tensor<string, []>("op_208_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6661184)))];
-            tensor<fp16, [1, 1500, 384]> v_5_cast = linear(bias = var_208_to_fp16, weight = var_207_to_fp16, x = var_188_cast);
-            tensor<int32, [4]> var_216 = const()[name = tensor<string, []>("op_216"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_217_cast = reshape(shape = var_216, x = q_5_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_30_to_fp16 = const()[name = tensor<string, []>("const_30_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 6, 64]> q_7_cast = mul(x = var_217_cast, y = const_30_to_fp16);
-            tensor<int32, [4]> var_223 = const()[name = tensor<string, []>("op_223"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_224_cast = reshape(shape = var_223, x = k_5_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_31_to_fp16 = const()[name = tensor<string, []>("const_31_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 6, 64]> k_7_cast = mul(x = var_224_cast, y = const_31_to_fp16);
-            tensor<int32, [4]> var_230 = const()[name = tensor<string, []>("op_230"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_231_cast = reshape(shape = var_230, x = v_5_cast);
-            tensor<int32, [4]> var_232 = const()[name = tensor<string, []>("op_232"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<bool, []> qk_3_transpose_x_0 = const()[name = tensor<string, []>("qk_3_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> qk_3_transpose_y_0 = const()[name = tensor<string, []>("qk_3_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<int32, [4]> transpose_10_perm_0 = const()[name = tensor<string, []>("transpose_10_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [4]> transpose_11_perm_0 = const()[name = tensor<string, []>("transpose_11_perm_0"), val = tensor<int32, [4]>([0, 2, 3, 1])];
-            tensor<fp16, [1, 6, 64, 1500]> transpose_25 = transpose(perm = transpose_11_perm_0, x = k_7_cast);
-            tensor<fp16, [1, 6, 1500, 64]> transpose_26 = transpose(perm = transpose_10_perm_0, x = q_7_cast);
-            tensor<fp16, [1, 6, 1500, 1500]> qk_3_cast = matmul(transpose_x = qk_3_transpose_x_0, transpose_y = qk_3_transpose_y_0, x = transpose_26, y = transpose_25);
-            tensor<fp16, [1, 6, 1500, 1500]> var_236_cast = softmax(axis = var_171, x = qk_3_cast);
-            tensor<bool, []> var_238_transpose_x_0 = const()[name = tensor<string, []>("op_238_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> var_238_transpose_y_0 = const()[name = tensor<string, []>("op_238_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [1, 6, 1500, 64]> transpose_27 = transpose(perm = var_232, x = var_231_cast);
-            tensor<fp16, [1, 6, 1500, 64]> var_238_cast = matmul(transpose_x = var_238_transpose_x_0, transpose_y = var_238_transpose_y_0, x = var_236_cast, y = transpose_27);
-            tensor<int32, [4]> var_239 = const()[name = tensor<string, []>("op_239"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [3]> concat_1 = const()[name = tensor<string, []>("concat_1"), val = tensor<int32, [3]>([1, 1500, 384])];
-            tensor<fp16, [1, 1500, 6, 64]> transpose_24 = transpose(perm = var_239, x = var_238_cast);
-            tensor<fp16, [1, 1500, 384]> x_23_cast = reshape(shape = concat_1, x = transpose_24);
-            tensor<fp16, [384, 384]> var_244_to_fp16 = const()[name = tensor<string, []>("op_244_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6662016)))];
-            tensor<fp16, [384]> var_245_to_fp16 = const()[name = tensor<string, []>("op_245_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6956992)))];
-            tensor<fp16, [1, 1500, 384]> var_246_cast = linear(bias = var_245_to_fp16, weight = var_244_to_fp16, x = x_23_cast);
-            tensor<fp16, [1, 1500, 384]> x_25_cast = add(x = x_19_cast, y = var_246_cast);
-            tensor<int32, [1]> var_252_axes_0 = const()[name = tensor<string, []>("op_252_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [384]> blocks_1_mlp_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_1_mlp_ln_weight_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6957824)))];
-            tensor<fp16, [384]> blocks_1_mlp_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_1_mlp_ln_bias_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6958656)))];
-            tensor<fp16, [1, 1500, 384]> var_252_cast = layer_norm(axes = var_252_axes_0, beta = blocks_1_mlp_ln_bias_to_fp16, epsilon = var_177_to_fp16, gamma = blocks_1_mlp_ln_weight_to_fp16, x = x_25_cast);
-            tensor<fp16, [1536, 384]> var_261_to_fp16 = const()[name = tensor<string, []>("op_261_to_fp16"), val = tensor<fp16, [1536, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6959488)))];
-            tensor<fp16, [1536]> var_262_to_fp16 = const()[name = tensor<string, []>("op_262_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(8139200)))];
-            tensor<fp16, [1, 1500, 1536]> input_17_cast = linear(bias = var_262_to_fp16, weight = var_261_to_fp16, x = var_252_cast);
-            tensor<string, []> x_29_mode_0 = const()[name = tensor<string, []>("x_29_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 1500, 1536]> x_29_cast = gelu(mode = x_29_mode_0, x = input_17_cast);
-            tensor<fp16, [384, 1536]> var_267_to_fp16 = const()[name = tensor<string, []>("op_267_to_fp16"), val = tensor<fp16, [384, 1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(8142336)))];
-            tensor<fp16, [384]> var_268_to_fp16 = const()[name = tensor<string, []>("op_268_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9322048)))];
-            tensor<fp16, [1, 1500, 384]> var_269_cast = linear(bias = var_268_to_fp16, weight = var_267_to_fp16, x = x_29_cast);
-            tensor<fp16, [1, 1500, 384]> x_31_cast = add(x = x_25_cast, y = var_269_cast);
-            tensor<int32, []> var_277 = const()[name = tensor<string, []>("op_277"), val = tensor<int32, []>(-1)];
-            tensor<int32, [1]> var_294_axes_0 = const()[name = tensor<string, []>("op_294_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [384]> blocks_2_attn_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_2_attn_ln_weight_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9322880)))];
-            tensor<fp16, [384]> blocks_2_attn_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_2_attn_ln_bias_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9323712)))];
-            tensor<fp16, []> var_283_to_fp16 = const()[name = tensor<string, []>("op_283_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 384]> var_294_cast = layer_norm(axes = var_294_axes_0, beta = blocks_2_attn_ln_bias_to_fp16, epsilon = var_283_to_fp16, gamma = blocks_2_attn_ln_weight_to_fp16, x = x_31_cast);
-            tensor<fp16, [384, 384]> var_305_to_fp16 = const()[name = tensor<string, []>("op_305_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9324544)))];
-            tensor<fp16, [384]> var_306_to_fp16 = const()[name = tensor<string, []>("op_306_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9619520)))];
-            tensor<fp16, [1, 1500, 384]> q_9_cast = linear(bias = var_306_to_fp16, weight = var_305_to_fp16, x = var_294_cast);
-            tensor<fp16, [384, 384]> var_309_to_fp16 = const()[name = tensor<string, []>("op_309_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9620352)))];
-            tensor<fp16, [384]> k_9_bias_0_to_fp16 = const()[name = tensor<string, []>("k_9_bias_0_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9915328)))];
-            tensor<fp16, [1, 1500, 384]> k_9_cast = linear(bias = k_9_bias_0_to_fp16, weight = var_309_to_fp16, x = var_294_cast);
-            tensor<fp16, [384, 384]> var_313_to_fp16 = const()[name = tensor<string, []>("op_313_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9916160)))];
-            tensor<fp16, [384]> var_314_to_fp16 = const()[name = tensor<string, []>("op_314_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10211136)))];
-            tensor<fp16, [1, 1500, 384]> v_9_cast = linear(bias = var_314_to_fp16, weight = var_313_to_fp16, x = var_294_cast);
-            tensor<int32, [4]> var_322 = const()[name = tensor<string, []>("op_322"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_323_cast = reshape(shape = var_322, x = q_9_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_32_to_fp16 = const()[name = tensor<string, []>("const_32_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 6, 64]> q_11_cast = mul(x = var_323_cast, y = const_32_to_fp16);
-            tensor<int32, [4]> var_329 = const()[name = tensor<string, []>("op_329"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_330_cast = reshape(shape = var_329, x = k_9_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_33_to_fp16 = const()[name = tensor<string, []>("const_33_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 6, 64]> k_11_cast = mul(x = var_330_cast, y = const_33_to_fp16);
-            tensor<int32, [4]> var_336 = const()[name = tensor<string, []>("op_336"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_337_cast = reshape(shape = var_336, x = v_9_cast);
-            tensor<int32, [4]> var_338 = const()[name = tensor<string, []>("op_338"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<bool, []> qk_5_transpose_x_0 = const()[name = tensor<string, []>("qk_5_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> qk_5_transpose_y_0 = const()[name = tensor<string, []>("qk_5_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<int32, [4]> transpose_12_perm_0 = const()[name = tensor<string, []>("transpose_12_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [4]> transpose_13_perm_0 = const()[name = tensor<string, []>("transpose_13_perm_0"), val = tensor<int32, [4]>([0, 2, 3, 1])];
-            tensor<fp16, [1, 6, 64, 1500]> transpose_21 = transpose(perm = transpose_13_perm_0, x = k_11_cast);
-            tensor<fp16, [1, 6, 1500, 64]> transpose_22 = transpose(perm = transpose_12_perm_0, x = q_11_cast);
-            tensor<fp16, [1, 6, 1500, 1500]> qk_5_cast = matmul(transpose_x = qk_5_transpose_x_0, transpose_y = qk_5_transpose_y_0, x = transpose_22, y = transpose_21);
-            tensor<fp16, [1, 6, 1500, 1500]> var_342_cast = softmax(axis = var_277, x = qk_5_cast);
-            tensor<bool, []> var_344_transpose_x_0 = const()[name = tensor<string, []>("op_344_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> var_344_transpose_y_0 = const()[name = tensor<string, []>("op_344_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [1, 6, 1500, 64]> transpose_23 = transpose(perm = var_338, x = var_337_cast);
-            tensor<fp16, [1, 6, 1500, 64]> var_344_cast = matmul(transpose_x = var_344_transpose_x_0, transpose_y = var_344_transpose_y_0, x = var_342_cast, y = transpose_23);
-            tensor<int32, [4]> var_345 = const()[name = tensor<string, []>("op_345"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [3]> concat_2 = const()[name = tensor<string, []>("concat_2"), val = tensor<int32, [3]>([1, 1500, 384])];
-            tensor<fp16, [1, 1500, 6, 64]> transpose_20 = transpose(perm = var_345, x = var_344_cast);
-            tensor<fp16, [1, 1500, 384]> x_35_cast = reshape(shape = concat_2, x = transpose_20);
-            tensor<fp16, [384, 384]> var_350_to_fp16 = const()[name = tensor<string, []>("op_350_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10211968)))];
-            tensor<fp16, [384]> var_351_to_fp16 = const()[name = tensor<string, []>("op_351_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10506944)))];
-            tensor<fp16, [1, 1500, 384]> var_352_cast = linear(bias = var_351_to_fp16, weight = var_350_to_fp16, x = x_35_cast);
-            tensor<fp16, [1, 1500, 384]> x_37_cast = add(x = x_31_cast, y = var_352_cast);
-            tensor<int32, [1]> var_358_axes_0 = const()[name = tensor<string, []>("op_358_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [384]> blocks_2_mlp_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_2_mlp_ln_weight_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10507776)))];
-            tensor<fp16, [384]> blocks_2_mlp_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_2_mlp_ln_bias_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10508608)))];
-            tensor<fp16, [1, 1500, 384]> var_358_cast = layer_norm(axes = var_358_axes_0, beta = blocks_2_mlp_ln_bias_to_fp16, epsilon = var_283_to_fp16, gamma = blocks_2_mlp_ln_weight_to_fp16, x = x_37_cast);
-            tensor<fp16, [1536, 384]> var_367_to_fp16 = const()[name = tensor<string, []>("op_367_to_fp16"), val = tensor<fp16, [1536, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10509440)))];
-            tensor<fp16, [1536]> var_368_to_fp16 = const()[name = tensor<string, []>("op_368_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(11689152)))];
-            tensor<fp16, [1, 1500, 1536]> input_25_cast = linear(bias = var_368_to_fp16, weight = var_367_to_fp16, x = var_358_cast);
-            tensor<string, []> x_41_mode_0 = const()[name = tensor<string, []>("x_41_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 1500, 1536]> x_41_cast = gelu(mode = x_41_mode_0, x = input_25_cast);
-            tensor<fp16, [384, 1536]> var_373_to_fp16 = const()[name = tensor<string, []>("op_373_to_fp16"), val = tensor<fp16, [384, 1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(11692288)))];
-            tensor<fp16, [384]> var_374_to_fp16 = const()[name = tensor<string, []>("op_374_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12872000)))];
-            tensor<fp16, [1, 1500, 384]> var_375_cast = linear(bias = var_374_to_fp16, weight = var_373_to_fp16, x = x_41_cast);
-            tensor<fp16, [1, 1500, 384]> x_43_cast = add(x = x_37_cast, y = var_375_cast);
-            tensor<int32, []> var_383 = const()[name = tensor<string, []>("op_383"), val = tensor<int32, []>(-1)];
-            tensor<int32, [1]> var_400_axes_0 = const()[name = tensor<string, []>("op_400_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [384]> blocks_3_attn_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_3_attn_ln_weight_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12872832)))];
-            tensor<fp16, [384]> blocks_3_attn_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_3_attn_ln_bias_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12873664)))];
-            tensor<fp16, []> var_389_to_fp16 = const()[name = tensor<string, []>("op_389_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 384]> var_400_cast = layer_norm(axes = var_400_axes_0, beta = blocks_3_attn_ln_bias_to_fp16, epsilon = var_389_to_fp16, gamma = blocks_3_attn_ln_weight_to_fp16, x = x_43_cast);
-            tensor<fp16, [384, 384]> var_411_to_fp16 = const()[name = tensor<string, []>("op_411_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12874496)))];
-            tensor<fp16, [384]> var_412_to_fp16 = const()[name = tensor<string, []>("op_412_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(13169472)))];
-            tensor<fp16, [1, 1500, 384]> q_13_cast = linear(bias = var_412_to_fp16, weight = var_411_to_fp16, x = var_400_cast);
-            tensor<fp16, [384, 384]> var_415_to_fp16 = const()[name = tensor<string, []>("op_415_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(13170304)))];
-            tensor<fp16, [384]> k_13_bias_0_to_fp16 = const()[name = tensor<string, []>("k_13_bias_0_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(13465280)))];
-            tensor<fp16, [1, 1500, 384]> k_13_cast = linear(bias = k_13_bias_0_to_fp16, weight = var_415_to_fp16, x = var_400_cast);
-            tensor<fp16, [384, 384]> var_419_to_fp16 = const()[name = tensor<string, []>("op_419_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(13466112)))];
-            tensor<fp16, [384]> var_420_to_fp16 = const()[name = tensor<string, []>("op_420_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(13761088)))];
-            tensor<fp16, [1, 1500, 384]> v_13_cast = linear(bias = var_420_to_fp16, weight = var_419_to_fp16, x = var_400_cast);
-            tensor<int32, [4]> var_428 = const()[name = tensor<string, []>("op_428"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_429_cast = reshape(shape = var_428, x = q_13_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_34_to_fp16 = const()[name = tensor<string, []>("const_34_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 6, 64]> q_cast = mul(x = var_429_cast, y = const_34_to_fp16);
-            tensor<int32, [4]> var_435 = const()[name = tensor<string, []>("op_435"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_436_cast = reshape(shape = var_435, x = k_13_cast);
-            tensor<fp16, [1, 1, 1, 1]> const_35_to_fp16 = const()[name = tensor<string, []>("const_35_to_fp16"), val = tensor<fp16, [1, 1, 1, 1]>([[[[0x1.6ap-2]]]])];
-            tensor<fp16, [1, 1500, 6, 64]> k_cast = mul(x = var_436_cast, y = const_35_to_fp16);
-            tensor<int32, [4]> var_442 = const()[name = tensor<string, []>("op_442"), val = tensor<int32, [4]>([1, 1500, 6, -1])];
-            tensor<fp16, [1, 1500, 6, 64]> var_443_cast = reshape(shape = var_442, x = v_13_cast);
-            tensor<int32, [4]> var_444 = const()[name = tensor<string, []>("op_444"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<bool, []> qk_transpose_x_0 = const()[name = tensor<string, []>("qk_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> qk_transpose_y_0 = const()[name = tensor<string, []>("qk_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<int32, [4]> transpose_14_perm_0 = const()[name = tensor<string, []>("transpose_14_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [4]> transpose_15_perm_0 = const()[name = tensor<string, []>("transpose_15_perm_0"), val = tensor<int32, [4]>([0, 2, 3, 1])];
-            tensor<fp16, [1, 6, 64, 1500]> transpose_17 = transpose(perm = transpose_15_perm_0, x = k_cast);
-            tensor<fp16, [1, 6, 1500, 64]> transpose_18 = transpose(perm = transpose_14_perm_0, x = q_cast);
-            tensor<fp16, [1, 6, 1500, 1500]> qk_cast = matmul(transpose_x = qk_transpose_x_0, transpose_y = qk_transpose_y_0, x = transpose_18, y = transpose_17);
-            tensor<fp16, [1, 6, 1500, 1500]> var_448_cast = softmax(axis = var_383, x = qk_cast);
-            tensor<bool, []> var_450_transpose_x_0 = const()[name = tensor<string, []>("op_450_transpose_x_0"), val = tensor<bool, []>(false)];
-            tensor<bool, []> var_450_transpose_y_0 = const()[name = tensor<string, []>("op_450_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [1, 6, 1500, 64]> transpose_19 = transpose(perm = var_444, x = var_443_cast);
-            tensor<fp16, [1, 6, 1500, 64]> var_450_cast = matmul(transpose_x = var_450_transpose_x_0, transpose_y = var_450_transpose_y_0, x = var_448_cast, y = transpose_19);
-            tensor<int32, [4]> var_451 = const()[name = tensor<string, []>("op_451"), val = tensor<int32, [4]>([0, 2, 1, 3])];
-            tensor<int32, [3]> concat_3 = const()[name = tensor<string, []>("concat_3"), val = tensor<int32, [3]>([1, 1500, 384])];
-            tensor<fp16, [1, 1500, 6, 64]> transpose_16 = transpose(perm = var_451, x = var_450_cast);
-            tensor<fp16, [1, 1500, 384]> x_47_cast = reshape(shape = concat_3, x = transpose_16);
-            tensor<fp16, [384, 384]> var_456_to_fp16 = const()[name = tensor<string, []>("op_456_to_fp16"), val = tensor<fp16, [384, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(13761920)))];
-            tensor<fp16, [384]> var_457_to_fp16 = const()[name = tensor<string, []>("op_457_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(14056896)))];
-            tensor<fp16, [1, 1500, 384]> var_458_cast = linear(bias = var_457_to_fp16, weight = var_456_to_fp16, x = x_47_cast);
-            tensor<fp16, [1, 1500, 384]> x_49_cast = add(x = x_43_cast, y = var_458_cast);
-            tensor<int32, [1]> var_464_axes_0 = const()[name = tensor<string, []>("op_464_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [384]> blocks_3_mlp_ln_weight_to_fp16 = const()[name = tensor<string, []>("blocks_3_mlp_ln_weight_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(14057728)))];
-            tensor<fp16, [384]> blocks_3_mlp_ln_bias_to_fp16 = const()[name = tensor<string, []>("blocks_3_mlp_ln_bias_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(14058560)))];
-            tensor<fp16, [1, 1500, 384]> var_464_cast = layer_norm(axes = var_464_axes_0, beta = blocks_3_mlp_ln_bias_to_fp16, epsilon = var_389_to_fp16, gamma = blocks_3_mlp_ln_weight_to_fp16, x = x_49_cast);
-            tensor<fp16, [1536, 384]> var_473_to_fp16 = const()[name = tensor<string, []>("op_473_to_fp16"), val = tensor<fp16, [1536, 384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(14059392)))];
-            tensor<fp16, [1536]> var_474_to_fp16 = const()[name = tensor<string, []>("op_474_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15239104)))];
-            tensor<fp16, [1, 1500, 1536]> input_33_cast = linear(bias = var_474_to_fp16, weight = var_473_to_fp16, x = var_464_cast);
-            tensor<string, []> x_53_mode_0 = const()[name = tensor<string, []>("x_53_mode_0"), val = tensor<string, []>("EXACT")];
-            tensor<fp16, [1, 1500, 1536]> x_53_cast = gelu(mode = x_53_mode_0, x = input_33_cast);
-            tensor<fp16, [384, 1536]> var_479_to_fp16 = const()[name = tensor<string, []>("op_479_to_fp16"), val = tensor<fp16, [384, 1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15242240)))];
-            tensor<fp16, [384]> var_480_to_fp16 = const()[name = tensor<string, []>("op_480_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(16421952)))];
-            tensor<fp16, [1, 1500, 384]> var_481_cast = linear(bias = var_480_to_fp16, weight = var_479_to_fp16, x = x_53_cast);
-            tensor<fp16, [1, 1500, 384]> x_cast = add(x = x_49_cast, y = var_481_cast);
-            tensor<int32, [1]> var_494_axes_0 = const()[name = tensor<string, []>("op_494_axes_0"), val = tensor<int32, [1]>([-1])];
-            tensor<fp16, [384]> ln_post_weight_to_fp16 = const()[name = tensor<string, []>("ln_post_weight_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(16422784)))];
-            tensor<fp16, [384]> ln_post_bias_to_fp16 = const()[name = tensor<string, []>("ln_post_bias_to_fp16"), val = tensor<fp16, [384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(16423616)))];
-            tensor<fp16, []> var_485_to_fp16 = const()[name = tensor<string, []>("op_485_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1500, 384]> var_494_cast = layer_norm(axes = var_494_axes_0, beta = ln_post_bias_to_fp16, epsilon = var_485_to_fp16, gamma = ln_post_weight_to_fp16, x = x_cast);
-            tensor<string, []> var_494_cast_to_fp32_dtype_0 = const()[name = tensor<string, []>("op_494_cast_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
-            tensor<fp32, [1, 1500, 384]> output = cast(dtype = var_494_cast_to_fp32_dtype_0, x = var_494_cast);
-        } -> (output);
-}

ggml-tiny-encoder.mlmodelc/weights/weight.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:dd51b56c3444d8e1ca678639f2ceaa03c883200304446549fcbbcedf63aa466f
-size 16424448