Supertonic 3 Core ML (FluidInference @ 2b5b939), arquivos usados pelo UXAR Reader

DurationPredictor_v2.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8bfa7996e1b4c0e26c9c1edc3861f28726bbd30f74d4a7519e69ef78c1b2d1c
+size 243

DurationPredictor_v2.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4f1630670632dd149787fb6ea4f934ab915c390a88d87a4e825ca58f041b4b2
+size 426

DurationPredictor_v2.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f709c544087966d6b2538ac50b7a4014d6e1049e21d713cf3bc71c4c5b9307c
+size 1797952

LICENSE

@@ -0,0 +1,217 @@
+Supertonic 3 (v1.7.3) — modelo de voz multilíngue
+Copyright (c) Supertone Inc. — https://huggingface.co/Supertone/supertonic-3
+Conversão para Core ML: FluidInference — https://huggingface.co/FluidInference/supertonic-3-coreml
+Os pesos do modelo são distribuídos sob a licença BigScience Open RAIL-M (texto completo abaixo),
+incluindo as restrições de uso do Anexo A, que se aplicam a quem usa o UXAR Reader.
+
+------------------------------------------------------------------------
+
+BigScience Open RAIL-M License
+dated August 18, 2022
+
+Section I: PREAMBLE
+
+This Open RAIL-M License was created by BigScience, a collaborative open innovation project aimed at
+the responsible development and use of large multilingual datasets and Large Language Models
+(“LLMs”). While a similar license was originally designed for the BLOOM model, we decided to adapt it
+and create this license in order to propose a general open and responsible license applicable to other
+machine learning based AI models (e.g. multimodal generative models).
+In short, this license strives for both the open and responsible downstream use of the accompanying
+model. When it comes to the open character, we took inspiration from open source permissive licenses
+regarding the grant of IP rights. Referring to the downstream responsible use, we added use-based
+restrictions not permitting the use of the Model in very specific scenarios, in order for the licensor to be
+able to enforce the license in case potential misuses of the Model may occur. Even though downstream
+derivative versions of the model could be released under different licensing terms, the latter will always
+have to include - at minimum - the same use-based restrictions as the ones in the original license (this
+license).
+The development and use of artificial intelligence (“AI”), does not come without concerns. The world has
+witnessed how AI techniques may, in some instances, become risky for the public in general. These risks
+come in many forms, from racial discrimination to the misuse of sensitive information.
+BigScience believes in the intersection between open and responsible AI development; thus, this License
+aims to strike a balance between both in order to enable responsible open-science in the field of AI.
+This License governs the use of the model (and its derivatives) and is informed by the model card
+associated with the model.
+
+NOW THEREFORE, You and Licensor agree as follows:
+
+1. Definitions
+(a) "License" means the terms and conditions for use, reproduction, and Distribution as defined in
+this document.
+(b) “Data” means a collection of information and/or content extracted from the dataset used with the
+Model, including to train, pretrain, or otherwise evaluate the Model. The Data is not licensed under
+this License.
+(c)“Output” means the results of operating a Model as embodied in informational content resulting
+therefrom.
+(d)“Model” means any accompanying machine-learning based assemblies (including checkpoints),
+consisting of learnt weights, parameters (including optimizer states), corresponding to the model
+architecture as embodied in the Complementary Material, that have been trained or tuned, in whole or
+in part on the Data, using the Complementary Material.
+(e) “Derivatives of the Model” means all modifications to the Model, works based on the Model, or any
+other model which is created or initialized by transfer of patterns of the weights, parameters,
+activations or output of the Model, to the other model, in order to cause the other model to perform
+similarly to the Model, including - but not limited to - distillation methods entailing the use of
+intermediate data representations or methods based on the generation of synthetic data by the Model
+for training the other model.
+(f)“Complementary Material” means the accompanying source code and scripts used to define,
+run, load, benchmark or evaluate the Model, and used to prepare data for training or evaluation, if
+any. This includes any accompanying documentation, tutorials, examples, etc, if any.
+(g) “Distribution” means any transmission, reproduction, publication or other sharing of the Model or
+Derivatives of the Model to a third party, including providing the Model as a hosted service made
+available by electronic or other remote means - e.g. API-based or web access.
+(h) “Licensor” means the copyright owner or entity authorized by the copyright owner that is
+granting the License, including the persons or entities that may have rights in the Model and/or
+distributing the Model.
+(i) "You" (or "Your") means an individual or Legal Entity exercising permissions granted by this
+License and/or making use of the Model for whichever purpose and in any field of use, including
+usage of the Model in an end-use application - e.g. chatbot, translator, image generator.
+(j) “Third Parties” means individuals or legal entities that are not under common control with
+Licensor or You.
+(k) "Contribution" means any work of authorship, including the original version of the Model and
+any modifications or additions to that Model or Derivatives of the Model thereof, that is
+intentionally submitted to Licensor for inclusion in the Model by the copyright owner or by an
+individual or Legal Entity authorized to submit on behalf of the copyright owner. For the
+purposes of this definition,
+“submitted” means any form of electronic, verbal, or written
+communication sent to the Licensor or its representatives, including but not limited to
+communication on electronic mailing lists, source code control systems, and issue tracking
+systems that are managed by, or on behalf of, the Licensor for the purpose of discussing and
+improving the Model, but excluding communication that is conspicuously marked or otherwise
+designated in writing by the copyright owner as "Not a Contribution."
+(l) "Contributor" means Licensor and any individual or Legal Entity on behalf of whom a
+Contribution has been received by Licensor and subsequently incorporated within the Model.
+
+
+Section II: INTELLECTUAL PROPERTY RIGHTS
+
+Both copyright and patent grants apply to the Model, Derivatives of the Model and Complementary
+Material. The Model and Derivatives of the Model are subject to additional terms as described in Section III.
+
+2. Grant of Copyright License. Subject to the terms and conditions of this License, each Contributor
+hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable copyright license to reproduce, prepare, publicly display, publicly perform, sublicense, and distribute the
+Complementary Material, the Model, and Derivatives of the Model.
+
+3. Grant of Patent License. Subject to the terms and conditions of this License and where and as
+applicable, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge,
+royalty-free, irrevocable (except as stated in this paragraph) patent license to make, have made, use, offer
+to sell, sell, import, and otherwise transfer the Model and the Complementary Material, where such
+license applies only to those patent claims licensable by such Contributor that are necessarily infringed by
+their Contribution(s) alone or by combination of their Contribution(s) with the Model to which such
+Contribution(s) was submitted. If You institute patent litigation against any entity (including a cross-claim
+or counterclaim in a lawsuit) alleging that the Model and/or Complementary Material or a Contribution
+incorporated within the Model and/or Complementary Material constitutes direct or contributory patent
+infringement, then any patent licenses granted to You under this License for the Model and/or Work shall
+terminate as of the date such litigation is asserted or filed.
+Section III: CONDITIONS OF USAGE, DISTRIBUTION AND REDISTRIBUTION
+
+4. Distribution and Redistribution. You may host for Third Party remote access purposes (e.g.
+software-as-a-service), reproduce and distribute copies of the Model or Derivatives of the Model thereof
+in any medium, with or without modifications, provided that You meet the following conditions:
+
+a. Use-based restrictions as referenced in paragraph 5 MUST be included as an enforceable provision
+by You in any type of legal agreement (e.g. a license) governing the use and/or distribution of the
+Model or Derivatives of the Model, and You shall give notice to subsequent users You Distribute to,
+that the Model or Derivatives of the Model are subject to paragraph 5. This provision does not apply
+to the use of Complementary Material.
+
+b. You must give any Third Party recipients of the Model or Derivatives of the Model a copy of this
+License;
+
+c. You must cause any modified files to carry prominent notices stating that You changed the files;
+
+d. You must retain all copyright, patent, trademark, and attribution notices excluding those notices
+that do not pertain to any part of the Model, Derivatives of the Model.
+You may add Your own copyright statement to Your modifications and may provide additional or
+different license terms and conditions - respecting paragraph 4.a.
+- for use, reproduction, or Distribution
+of Your modifications, or for any such Derivatives of the Model as a whole, provided Your use,
+reproduction, and Distribution of the Model otherwise complies with the conditions stated in this License.
+
+5. Use-based restrictions. The restrictions set forth in Attachment A are considered Use-based restrictions.
+Therefore You cannot use the Model and the Derivatives of the Model for the specified restricted uses. You
+may use the Model subject to this License, including only for lawful purposes and in accordance with the
+License. Use may include creating any content with, finetuning, updating, running, training, evaluating and/or
+reparametrizing the Model. You shall require all of Your users who use the Model or a Derivative of the Model
+to comply with the terms of this paragraph (paragraph 5).
+
+6. The Output You Generate. Except as set forth herein, Licensor claims no rights in the Output You
+generate using the Model. You are accountable for the Output you generate and its subsequent uses. No
+use of the output can contravene any provision as stated in the License.
+
+Section IV: OTHER PROVISIONS
+
+7. Updates and Runtime Restrictions. To the maximum extent permitted by law, Licensor reserves the
+right to restrict (remotely or otherwise) usage of the Model in violation of this License, update the Model
+through electronic means, or modify the Output of the Model based on updates. You shall undertake
+reasonable efforts to use the latest version of the Model.
+
+8. Trademarks and related. Nothing in this License permits You to make use of Licensors’ trademarks,
+trade names, logos or to otherwise suggest endorsement or misrepresent the relationship between the
+parties; and any rights not expressly granted herein are reserved by the Licensors.
+
+9. Disclaimer of Warranty. Unless required by applicable law or agreed to in writing, Licensor provides
+the Model and the Complementary Material (and each Contributor provides its Contributions) on an "AS
+IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied,
+including, without limitation, any warranties or conditions of TITLE, NON-INFRINGEMENT,
+MERCHANTABILITY , or FITNESS FOR A PARTICULAR PURPOSE. You are solely responsible for
+determining the appropriateness of using or redistributing the Model, Derivatives of the Model, and the
+Complementary Material and assume any risks associated with Your exercise of permissions under this
+License.
+
+10. Limitation of Liability. In no event and under no legal theory, whether in tort (including negligence),
+contract, or otherwise, unless required by applicable law (such as deliberate and grossly negligent acts) or
+agreed to in writing, shall any Contributor be liable to You for damages, including any direct, indirect,
+special, incidental, or consequential damages of any character arising as a result of this License or out of
+the use or inability to use the Model and the Complementary Material (including but not limited to
+damages for loss of goodwill, work stoppage, computer failure or malfunction, or any and all other
+commercial damages or losses), even if such Contributor has been advised of the possibility of such
+damages.
+
+11. Accepting Warranty or Additional Liability. While redistributing the Model, Derivatives of the
+Model and the Complementary Material thereof, You may choose to offer, and charge a fee for, acceptance
+of support, warranty, indemnity, or other liability obligations and/or rights consistent with this License.
+However, in accepting such obligations, You may act only on Your own behalf and on Your sole
+responsibility, not on behalf of any other Contributor, and only if You agree to indemnify, defend, and
+hold each Contributor harmless for any liability incurred by, or claims asserted against, such Contributor
+by reason of your accepting any such warranty or additional liability.
+
+12. If any provision of this License is held to be invalid, illegal or unenforceable, the remaining
+provisions shall be unaffected thereby and remain valid as if such provision had not been set forth herein.
+
+END OF TERMS AND CONDITIONS
+
+Attachment A
+
+Use Restrictions
+
+You agree not to use the Model or Derivatives of the Model:
+(a) In any way that violates any applicable national, federal, state, local or international law
+or regulation;
+(b) For the purpose of exploiting, harming or attempting to exploit or harm minors in any
+way;
+(c) To generate or disseminate verifiably false information and/or content with the purpose of
+harming others;
+(d) To generate or disseminate personal identifiable information that can be used to harm an
+individual;
+(e) To generate or disseminate information and/or content (e.g. images, code, posts, articles),
+and place the information and/or content in any context (e.g. bot generating tweets)
+without expressly and intelligibly disclaiming that the information and/or content is
+machine generated;
+(f) To defame, disparage or otherwise harass others;
+(g) To impersonate or attempt to impersonate (e.g. deepfakes) others without their consent;
+(h) For fully automated decision making that adversely impacts an individual’s legal rights or
+otherwise creates or modifies a binding, enforceable obligation;
+(i) For any use intended to or which has the effect of discriminating against or harming
+individuals or groups based on online or offline social behavior or known or predicted
+personal or personality characteristics;
+(j) To exploit any of the vulnerabilities of a specific group of persons based on their age,
+social, physical or mental characteristics, in order to materially distort the behavior of a
+person pertaining to that group in a manner that causes or is likely to cause that person or
+another person physical or psychological harm;
+(k) For any use intended to or which has the effect of discriminating against individuals or
+groups based on legally protected characteristics or categories;
+(l) To provide medical advice and medical results interpretation;
+(m) To generate or disseminate information for the purpose to be used for administration of
+justice, law enforcement, immigration or asylum processes, such as predicting an
+individual will commit fraud/crime commitment (e.g. by text profiling, drawing causal
+relationships between assertions made in documents, indiscriminate and
+arbitrarily-targeted use).

README.md

@@ -1,3 +1,23 @@
 ---
 license: bigscience-openrail-m
+base_model: FluidInference/supertonic-3-coreml
+tags:
+  - text-to-speech
+  - coreml
 ---
+
+# Supertonic 3 (Core ML) para iOS (UXAR Reader)
+
+Cópia dos arquivos do modelo de voz **Supertonic 3** em Core ML, usados pelas vozes
+naturais em vários idiomas do app UXAR Reader.
+
+- Modelo original: Supertonic 3, por Supertone Inc.
+- Conversão para Core ML: [FluidInference/supertonic-3-coreml](https://huggingface.co/FluidInference/supertonic-3-coreml),
+  commit `2b5b939dfa1bca362c5212dcafeb8aadde90b1bc`
+- Licença: OpenRAIL-M (a mesma do original) — texto completo, com as restrições de uso
+  (Anexo A), no arquivo `LICENSE`. Quem usar ou redistribuir estes arquivos fica sujeito
+  às mesmas restrições de uso.
+
+Só os 28 arquivos que o app usa (versão v2, VectorEstimator int8, vocoder e as 10 vozes).
+Nenhum arquivo foi alterado; tamanho e SHA-256 de cada um estão em `files.txt`
+(caminho, bytes, SHA-256) e são conferidos pelo app a cada download.

TextEncoder_v2.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:973b90afda5d837c037cef2dcbbda275f005eeeb55f2712ca6fbcfa68220f5b5
+size 243

TextEncoder_v2.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1177e3e03cc4130e5dca18eb007f20573bf95d3938abf62ab2d3470b04cbe7f
+size 432

TextEncoder_v2.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c7afa5d02426a8a363c9cef0e27ce64a5c57a67d1a59f60aac0ac58ffb9877f
+size 18022016

VectorEstimatorVariants/VectorEstimator_v2_int8.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d27fc7601864cc7e4ee95a4d33f7202a014fcf548fd89f3a50fe1cf83900c1e5
+size 243

VectorEstimatorVariants/VectorEstimator_v2_int8.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a61ba6647e635b19ba1bbc39d4e3fa35ad5c18bef4f9513c3137a2a0d4a6705
+size 634

VectorEstimatorVariants/VectorEstimator_v2_int8.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e637c3a1725b5e1b3df09cc162d1822e8fe80fa92ba01c180dde2ca29c12523
+size 64184608

Vocoder.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4fe1b825137629a96dc58a1339bc4ece32041b755f99d638f21a153f2e7faed6
+size 243

Vocoder.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbf54fc98879ee1c9a213452e16d09367f3c34ad489b175bcae7ef5f74f8d121
+size 388

Vocoder.mlmodelc/model.mil

@@ -0,0 +1,465 @@
+program(1.3)
+[buildInfo = dict<string, string>({{"coremlc-component-MIL", "3520.4.1"}, {"coremlc-version", "3520.5.1"}, {"coremltools-component-torch", "2.5.1"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "9.0"}})]
+{
+    func main<ios18>(tensor<fp32, [1, 144, ?]> latent) [FlexibleShapeInformation = tuple<tuple<string, dict<string, tensor<int32, [?]>>>, tuple<string, dict<string, list<tensor<int32, [2]>, ?>>>>((("DefaultShapes", {{"latent", [1, 144, 4]}}), ("RangeDims", {{"latent", [[1, 1], [144, 144], [4, 512]]}})))] {
+            string latent_to_fp16_dtype_0 = const()[name = string("latent_to_fp16_dtype_0"), val = string("fp16")];
+            fp16 _inversed_x_1_y_0_to_fp16 = const()[name = string("_inversed_x_1_y_0_to_fp16"), val = fp16(0x1p+2)];
+            tensor<fp16, [1, 144, ?]> latent_to_fp16 = cast(dtype = latent_to_fp16_dtype_0, x = latent)[name = string("cast_6")];
+            tensor<fp16, [1, 144, ?]> _inversed_x_1_cast_fp16 = mul(x = latent_to_fp16, y = _inversed_x_1_y_0_to_fp16)[name = string("_inversed_x_1_cast_fp16")];
+            tensor<int32, [4]> var_33 = const()[name = string("op_33"), val = tensor<int32, [4]>([1, 24, 6, -1])];
+            tensor<fp16, [1, 24, 6, ?]> x_3_cast_fp16 = reshape(shape = var_33, x = _inversed_x_1_cast_fp16)[name = string("x_3_cast_fp16")];
+            tensor<int32, [4]> var_37_perm_0 = const()[name = string("op_37_perm_0"), val = tensor<int32, [4]>([0, 1, 3, 2])];
+            tensor<int32, [3]> var_43 = const()[name = string("op_43"), val = tensor<int32, [3]>([1, 24, -1])];
+            tensor<fp16, [1, 24, ?, 6]> var_37_cast_fp16 = transpose(perm = var_37_perm_0, x = x_3_cast_fp16)[name = string("transpose_21")];
+            tensor<fp16, [1, 24, ?]> x_7_cast_fp16 = reshape(shape = var_43, x = var_37_cast_fp16)[name = string("x_7_cast_fp16")];
+            tensor<fp16, [1, 24, 1]> latent_std_to_fp16 = const()[name = string("latent_std_to_fp16"), val = tensor<fp16, [1, 24, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(64)))];
+            tensor<fp16, [1, 24, ?]> var_45_cast_fp16 = mul(x = x_7_cast_fp16, y = latent_std_to_fp16)[name = string("op_45_cast_fp16")];
+            tensor<fp16, [1, 24, 1]> latent_mean_to_fp16 = const()[name = string("latent_mean_to_fp16"), val = tensor<fp16, [1, 24, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(192)))];
+            tensor<fp16, [1, 24, ?]> input_1_cast_fp16 = add(x = var_45_cast_fp16, y = latent_mean_to_fp16)[name = string("input_1_cast_fp16")];
+            tensor<int32, [6]> input_3_pad_0 = const()[name = string("input_3_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 6, 0])];
+            string input_3_mode_0 = const()[name = string("input_3_mode_0"), val = string("replicate")];
+            fp16 const_0_to_fp16 = const()[name = string("const_0_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 24, ?]> input_3_cast_fp16 = pad(constant_val = const_0_to_fp16, mode = input_3_mode_0, pad = input_3_pad_0, x = input_1_cast_fp16)[name = string("input_3_cast_fp16")];
+            string input_5_pad_type_0 = const()[name = string("input_5_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> input_5_strides_0 = const()[name = string("input_5_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> input_5_pad_0 = const()[name = string("input_5_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> input_5_dilations_0 = const()[name = string("input_5_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 input_5_groups_0 = const()[name = string("input_5_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 24, 7]> embed_net_weight_to_fp16 = const()[name = string("embed_net_weight_to_fp16"), val = tensor<fp16, [512, 24, 7]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(320)))];
+            tensor<fp16, [512]> embed_net_bias_to_fp16 = const()[name = string("embed_net_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(172416)))];
+            tensor<fp16, [1, 512, ?]> input_5_cast_fp16 = conv(bias = embed_net_bias_to_fp16, dilations = input_5_dilations_0, groups = input_5_groups_0, pad = input_5_pad_0, pad_type = input_5_pad_type_0, strides = input_5_strides_0, weight = embed_net_weight_to_fp16, x = input_3_cast_fp16)[name = string("input_5_cast_fp16")];
+            tensor<int32, [6]> input_7_pad_0 = const()[name = string("input_7_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 6, 0])];
+            string input_7_mode_0 = const()[name = string("input_7_mode_0"), val = string("replicate")];
+            fp16 const_1_to_fp16 = const()[name = string("const_1_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 512, ?]> input_7_cast_fp16 = pad(constant_val = const_1_to_fp16, mode = input_7_mode_0, pad = input_7_pad_0, x = input_5_cast_fp16)[name = string("input_7_cast_fp16")];
+            string x_9_pad_type_0 = const()[name = string("x_9_pad_type_0"), val = string("valid")];
+            int32 x_9_groups_0 = const()[name = string("x_9_groups_0"), val = int32(512)];
+            tensor<int32, [1]> x_9_strides_0 = const()[name = string("x_9_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> x_9_pad_0 = const()[name = string("x_9_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> x_9_dilations_0 = const()[name = string("x_9_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<fp16, [512, 1, 7]> convnext_0_dwconv_net_weight_to_fp16 = const()[name = string("convnext_0_dwconv_net_weight_to_fp16"), val = tensor<fp16, [512, 1, 7]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(173504)))];
+            tensor<fp16, [512]> convnext_0_dwconv_net_bias_to_fp16 = const()[name = string("convnext_0_dwconv_net_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(180736)))];
+            tensor<fp16, [1, 512, ?]> x_9_cast_fp16 = conv(bias = convnext_0_dwconv_net_bias_to_fp16, dilations = x_9_dilations_0, groups = x_9_groups_0, pad = x_9_pad_0, pad_type = x_9_pad_type_0, strides = x_9_strides_0, weight = convnext_0_dwconv_net_weight_to_fp16, x = input_7_cast_fp16)[name = string("x_9_cast_fp16")];
+            tensor<int32, [3]> input_9_perm_0 = const()[name = string("input_9_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [1]> var_96_axes_0 = const()[name = string("op_96_axes_0"), val = tensor<int32, [1]>([-1])];
+            tensor<fp16, [512]> convnext_0_norm_norm_weight_to_fp16 = const()[name = string("convnext_0_norm_norm_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(181824)))];
+            tensor<fp16, [512]> convnext_0_norm_norm_bias_to_fp16 = const()[name = string("convnext_0_norm_norm_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(182912)))];
+            fp16 var_67_to_fp16 = const()[name = string("op_67_to_fp16"), val = fp16(0x1.5p-17)];
+            tensor<fp16, [1, ?, 512]> input_9_cast_fp16 = transpose(perm = input_9_perm_0, x = x_9_cast_fp16)[name = string("transpose_20")];
+            tensor<fp16, [1, ?, 512]> var_96_cast_fp16 = layer_norm(axes = var_96_axes_0, beta = convnext_0_norm_norm_bias_to_fp16, epsilon = var_67_to_fp16, gamma = convnext_0_norm_norm_weight_to_fp16, x = input_9_cast_fp16)[name = string("op_96_cast_fp16")];
+            tensor<int32, [3]> input_11_perm_0 = const()[name = string("input_11_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            string h_1_pad_type_0 = const()[name = string("h_1_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_1_strides_0 = const()[name = string("h_1_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_1_pad_0 = const()[name = string("h_1_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_1_dilations_0 = const()[name = string("h_1_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_1_groups_0 = const()[name = string("h_1_groups_0"), val = int32(1)];
+            tensor<fp16, [2048, 512, 1]> convnext_0_pwconv1_weight_to_fp16 = const()[name = string("convnext_0_pwconv1_weight_to_fp16"), val = tensor<fp16, [2048, 512, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(184000)))];
+            tensor<fp16, [2048]> convnext_0_pwconv1_bias_to_fp16 = const()[name = string("convnext_0_pwconv1_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(2281216)))];
+            tensor<fp16, [1, 512, ?]> input_11_cast_fp16 = transpose(perm = input_11_perm_0, x = var_96_cast_fp16)[name = string("transpose_19")];
+            tensor<fp16, [1, 2048, ?]> h_1_cast_fp16 = conv(bias = convnext_0_pwconv1_bias_to_fp16, dilations = h_1_dilations_0, groups = h_1_groups_0, pad = h_1_pad_0, pad_type = h_1_pad_type_0, strides = h_1_strides_0, weight = convnext_0_pwconv1_weight_to_fp16, x = input_11_cast_fp16)[name = string("h_1_cast_fp16")];
+            string input_13_mode_0 = const()[name = string("input_13_mode_0"), val = string("EXACT")];
+            tensor<fp16, [1, 2048, ?]> input_13_cast_fp16 = gelu(mode = input_13_mode_0, x = h_1_cast_fp16)[name = string("input_13_cast_fp16")];
+            string h_3_pad_type_0 = const()[name = string("h_3_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_3_strides_0 = const()[name = string("h_3_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_3_pad_0 = const()[name = string("h_3_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_3_dilations_0 = const()[name = string("h_3_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_3_groups_0 = const()[name = string("h_3_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 2048, 1]> var_113_weight_0_to_fp16 = const()[name = string("op_113_weight_0_to_fp16"), val = tensor<fp16, [512, 2048, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(2285376)))];
+            tensor<fp16, [512]> var_113_bias_0_to_fp16 = const()[name = string("op_113_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(4382592)))];
+            tensor<fp16, [1, 512, ?]> var_113_cast_fp16 = conv(bias = var_113_bias_0_to_fp16, dilations = h_3_dilations_0, groups = h_3_groups_0, pad = h_3_pad_0, pad_type = h_3_pad_type_0, strides = h_3_strides_0, weight = var_113_weight_0_to_fp16, x = input_13_cast_fp16)[name = string("op_113_cast_fp16")];
+            tensor<fp16, [1, 512, ?]> input_15_cast_fp16 = add(x = input_5_cast_fp16, y = var_113_cast_fp16)[name = string("input_15_cast_fp16")];
+            tensor<int32, [6]> input_17_pad_0 = const()[name = string("input_17_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 12, 0])];
+            string input_17_mode_0 = const()[name = string("input_17_mode_0"), val = string("replicate")];
+            fp16 const_2_to_fp16 = const()[name = string("const_2_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 512, ?]> input_17_cast_fp16 = pad(constant_val = const_2_to_fp16, mode = input_17_mode_0, pad = input_17_pad_0, x = input_15_cast_fp16)[name = string("input_17_cast_fp16")];
+            string x_11_pad_type_0 = const()[name = string("x_11_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> x_11_dilations_0 = const()[name = string("x_11_dilations_0"), val = tensor<int32, [1]>([2])];
+            int32 x_11_groups_0 = const()[name = string("x_11_groups_0"), val = int32(512)];
+            tensor<int32, [1]> x_11_strides_0 = const()[name = string("x_11_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> x_11_pad_0 = const()[name = string("x_11_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<fp16, [512, 1, 7]> convnext_1_dwconv_net_weight_to_fp16 = const()[name = string("convnext_1_dwconv_net_weight_to_fp16"), val = tensor<fp16, [512, 1, 7]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(4383680)))];
+            tensor<fp16, [512]> convnext_1_dwconv_net_bias_to_fp16 = const()[name = string("convnext_1_dwconv_net_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(4390912)))];
+            tensor<fp16, [1, 512, ?]> x_11_cast_fp16 = conv(bias = convnext_1_dwconv_net_bias_to_fp16, dilations = x_11_dilations_0, groups = x_11_groups_0, pad = x_11_pad_0, pad_type = x_11_pad_type_0, strides = x_11_strides_0, weight = convnext_1_dwconv_net_weight_to_fp16, x = input_17_cast_fp16)[name = string("x_11_cast_fp16")];
+            tensor<int32, [3]> input_19_perm_0 = const()[name = string("input_19_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [1]> var_146_axes_0 = const()[name = string("op_146_axes_0"), val = tensor<int32, [1]>([-1])];
+            tensor<fp16, [512]> convnext_1_norm_norm_weight_to_fp16 = const()[name = string("convnext_1_norm_norm_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(4392000)))];
+            tensor<fp16, [512]> convnext_1_norm_norm_bias_to_fp16 = const()[name = string("convnext_1_norm_norm_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(4393088)))];
+            fp16 var_116_to_fp16 = const()[name = string("op_116_to_fp16"), val = fp16(0x1.5p-17)];
+            tensor<fp16, [1, ?, 512]> input_19_cast_fp16 = transpose(perm = input_19_perm_0, x = x_11_cast_fp16)[name = string("transpose_18")];
+            tensor<fp16, [1, ?, 512]> var_146_cast_fp16 = layer_norm(axes = var_146_axes_0, beta = convnext_1_norm_norm_bias_to_fp16, epsilon = var_116_to_fp16, gamma = convnext_1_norm_norm_weight_to_fp16, x = input_19_cast_fp16)[name = string("op_146_cast_fp16")];
+            tensor<int32, [3]> input_21_perm_0 = const()[name = string("input_21_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            string h_5_pad_type_0 = const()[name = string("h_5_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_5_strides_0 = const()[name = string("h_5_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_5_pad_0 = const()[name = string("h_5_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_5_dilations_0 = const()[name = string("h_5_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_5_groups_0 = const()[name = string("h_5_groups_0"), val = int32(1)];
+            tensor<fp16, [2048, 512, 1]> convnext_1_pwconv1_weight_to_fp16 = const()[name = string("convnext_1_pwconv1_weight_to_fp16"), val = tensor<fp16, [2048, 512, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(4394176)))];
+            tensor<fp16, [2048]> convnext_1_pwconv1_bias_to_fp16 = const()[name = string("convnext_1_pwconv1_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(6491392)))];
+            tensor<fp16, [1, 512, ?]> input_21_cast_fp16 = transpose(perm = input_21_perm_0, x = var_146_cast_fp16)[name = string("transpose_17")];
+            tensor<fp16, [1, 2048, ?]> h_5_cast_fp16 = conv(bias = convnext_1_pwconv1_bias_to_fp16, dilations = h_5_dilations_0, groups = h_5_groups_0, pad = h_5_pad_0, pad_type = h_5_pad_type_0, strides = h_5_strides_0, weight = convnext_1_pwconv1_weight_to_fp16, x = input_21_cast_fp16)[name = string("h_5_cast_fp16")];
+            string input_23_mode_0 = const()[name = string("input_23_mode_0"), val = string("EXACT")];
+            tensor<fp16, [1, 2048, ?]> input_23_cast_fp16 = gelu(mode = input_23_mode_0, x = h_5_cast_fp16)[name = string("input_23_cast_fp16")];
+            string h_7_pad_type_0 = const()[name = string("h_7_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_7_strides_0 = const()[name = string("h_7_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_7_pad_0 = const()[name = string("h_7_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_7_dilations_0 = const()[name = string("h_7_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_7_groups_0 = const()[name = string("h_7_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 2048, 1]> var_163_weight_0_to_fp16 = const()[name = string("op_163_weight_0_to_fp16"), val = tensor<fp16, [512, 2048, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(6495552)))];
+            tensor<fp16, [512]> var_163_bias_0_to_fp16 = const()[name = string("op_163_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(8592768)))];
+            tensor<fp16, [1, 512, ?]> var_163_cast_fp16 = conv(bias = var_163_bias_0_to_fp16, dilations = h_7_dilations_0, groups = h_7_groups_0, pad = h_7_pad_0, pad_type = h_7_pad_type_0, strides = h_7_strides_0, weight = var_163_weight_0_to_fp16, x = input_23_cast_fp16)[name = string("op_163_cast_fp16")];
+            tensor<fp16, [1, 512, ?]> input_25_cast_fp16 = add(x = input_15_cast_fp16, y = var_163_cast_fp16)[name = string("input_25_cast_fp16")];
+            tensor<int32, [6]> input_27_pad_0 = const()[name = string("input_27_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 24, 0])];
+            string input_27_mode_0 = const()[name = string("input_27_mode_0"), val = string("replicate")];
+            fp16 const_3_to_fp16 = const()[name = string("const_3_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 512, ?]> input_27_cast_fp16 = pad(constant_val = const_3_to_fp16, mode = input_27_mode_0, pad = input_27_pad_0, x = input_25_cast_fp16)[name = string("input_27_cast_fp16")];
+            string x_13_pad_type_0 = const()[name = string("x_13_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> x_13_dilations_0 = const()[name = string("x_13_dilations_0"), val = tensor<int32, [1]>([4])];
+            int32 x_13_groups_0 = const()[name = string("x_13_groups_0"), val = int32(512)];
+            tensor<int32, [1]> x_13_strides_0 = const()[name = string("x_13_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> x_13_pad_0 = const()[name = string("x_13_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<fp16, [512, 1, 7]> convnext_2_dwconv_net_weight_to_fp16 = const()[name = string("convnext_2_dwconv_net_weight_to_fp16"), val = tensor<fp16, [512, 1, 7]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(8593856)))];
+            tensor<fp16, [512]> convnext_2_dwconv_net_bias_to_fp16 = const()[name = string("convnext_2_dwconv_net_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(8601088)))];
+            tensor<fp16, [1, 512, ?]> x_13_cast_fp16 = conv(bias = convnext_2_dwconv_net_bias_to_fp16, dilations = x_13_dilations_0, groups = x_13_groups_0, pad = x_13_pad_0, pad_type = x_13_pad_type_0, strides = x_13_strides_0, weight = convnext_2_dwconv_net_weight_to_fp16, x = input_27_cast_fp16)[name = string("x_13_cast_fp16")];
+            tensor<int32, [3]> input_29_perm_0 = const()[name = string("input_29_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [1]> var_197_axes_0 = const()[name = string("op_197_axes_0"), val = tensor<int32, [1]>([-1])];
+            tensor<fp16, [512]> convnext_2_norm_norm_weight_to_fp16 = const()[name = string("convnext_2_norm_norm_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(8602176)))];
+            tensor<fp16, [512]> convnext_2_norm_norm_bias_to_fp16 = const()[name = string("convnext_2_norm_norm_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(8603264)))];
+            fp16 var_167_to_fp16 = const()[name = string("op_167_to_fp16"), val = fp16(0x1.5p-17)];
+            tensor<fp16, [1, ?, 512]> input_29_cast_fp16 = transpose(perm = input_29_perm_0, x = x_13_cast_fp16)[name = string("transpose_16")];
+            tensor<fp16, [1, ?, 512]> var_197_cast_fp16 = layer_norm(axes = var_197_axes_0, beta = convnext_2_norm_norm_bias_to_fp16, epsilon = var_167_to_fp16, gamma = convnext_2_norm_norm_weight_to_fp16, x = input_29_cast_fp16)[name = string("op_197_cast_fp16")];
+            tensor<int32, [3]> input_31_perm_0 = const()[name = string("input_31_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            string h_9_pad_type_0 = const()[name = string("h_9_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_9_strides_0 = const()[name = string("h_9_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_9_pad_0 = const()[name = string("h_9_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_9_dilations_0 = const()[name = string("h_9_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_9_groups_0 = const()[name = string("h_9_groups_0"), val = int32(1)];
+            tensor<fp16, [2048, 512, 1]> convnext_2_pwconv1_weight_to_fp16 = const()[name = string("convnext_2_pwconv1_weight_to_fp16"), val = tensor<fp16, [2048, 512, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(8604352)))];
+            tensor<fp16, [2048]> convnext_2_pwconv1_bias_to_fp16 = const()[name = string("convnext_2_pwconv1_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(10701568)))];
+            tensor<fp16, [1, 512, ?]> input_31_cast_fp16 = transpose(perm = input_31_perm_0, x = var_197_cast_fp16)[name = string("transpose_15")];
+            tensor<fp16, [1, 2048, ?]> h_9_cast_fp16 = conv(bias = convnext_2_pwconv1_bias_to_fp16, dilations = h_9_dilations_0, groups = h_9_groups_0, pad = h_9_pad_0, pad_type = h_9_pad_type_0, strides = h_9_strides_0, weight = convnext_2_pwconv1_weight_to_fp16, x = input_31_cast_fp16)[name = string("h_9_cast_fp16")];
+            string input_33_mode_0 = const()[name = string("input_33_mode_0"), val = string("EXACT")];
+            tensor<fp16, [1, 2048, ?]> input_33_cast_fp16 = gelu(mode = input_33_mode_0, x = h_9_cast_fp16)[name = string("input_33_cast_fp16")];
+            string h_11_pad_type_0 = const()[name = string("h_11_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_11_strides_0 = const()[name = string("h_11_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_11_pad_0 = const()[name = string("h_11_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_11_dilations_0 = const()[name = string("h_11_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_11_groups_0 = const()[name = string("h_11_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 2048, 1]> var_214_weight_0_to_fp16 = const()[name = string("op_214_weight_0_to_fp16"), val = tensor<fp16, [512, 2048, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(10705728)))];
+            tensor<fp16, [512]> var_214_bias_0_to_fp16 = const()[name = string("op_214_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(12802944)))];
+            tensor<fp16, [1, 512, ?]> var_214_cast_fp16 = conv(bias = var_214_bias_0_to_fp16, dilations = h_11_dilations_0, groups = h_11_groups_0, pad = h_11_pad_0, pad_type = h_11_pad_type_0, strides = h_11_strides_0, weight = var_214_weight_0_to_fp16, x = input_33_cast_fp16)[name = string("op_214_cast_fp16")];
+            tensor<fp16, [1, 512, ?]> input_35_cast_fp16 = add(x = input_25_cast_fp16, y = var_214_cast_fp16)[name = string("input_35_cast_fp16")];
+            tensor<int32, [6]> input_37_pad_0 = const()[name = string("input_37_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 6, 0])];
+            string input_37_mode_0 = const()[name = string("input_37_mode_0"), val = string("replicate")];
+            fp16 const_4_to_fp16 = const()[name = string("const_4_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 512, ?]> input_37_cast_fp16 = pad(constant_val = const_4_to_fp16, mode = input_37_mode_0, pad = input_37_pad_0, x = input_35_cast_fp16)[name = string("input_37_cast_fp16")];
+            string x_15_pad_type_0 = const()[name = string("x_15_pad_type_0"), val = string("valid")];
+            int32 x_15_groups_0 = const()[name = string("x_15_groups_0"), val = int32(512)];
+            tensor<int32, [1]> x_15_strides_0 = const()[name = string("x_15_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> x_15_pad_0 = const()[name = string("x_15_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> x_15_dilations_0 = const()[name = string("x_15_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<fp16, [512, 1, 7]> convnext_3_dwconv_net_weight_to_fp16 = const()[name = string("convnext_3_dwconv_net_weight_to_fp16"), val = tensor<fp16, [512, 1, 7]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(12804032)))];
+            tensor<fp16, [512]> convnext_3_dwconv_net_bias_to_fp16 = const()[name = string("convnext_3_dwconv_net_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(12811264)))];
+            tensor<fp16, [1, 512, ?]> x_15_cast_fp16 = conv(bias = convnext_3_dwconv_net_bias_to_fp16, dilations = x_15_dilations_0, groups = x_15_groups_0, pad = x_15_pad_0, pad_type = x_15_pad_type_0, strides = x_15_strides_0, weight = convnext_3_dwconv_net_weight_to_fp16, x = input_37_cast_fp16)[name = string("x_15_cast_fp16")];
+            tensor<int32, [3]> input_39_perm_0 = const()[name = string("input_39_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [1]> var_247_axes_0 = const()[name = string("op_247_axes_0"), val = tensor<int32, [1]>([-1])];
+            tensor<fp16, [512]> convnext_3_norm_norm_weight_to_fp16 = const()[name = string("convnext_3_norm_norm_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(12812352)))];
+            tensor<fp16, [512]> convnext_3_norm_norm_bias_to_fp16 = const()[name = string("convnext_3_norm_norm_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(12813440)))];
+            fp16 var_218_to_fp16 = const()[name = string("op_218_to_fp16"), val = fp16(0x1.5p-17)];
+            tensor<fp16, [1, ?, 512]> input_39_cast_fp16 = transpose(perm = input_39_perm_0, x = x_15_cast_fp16)[name = string("transpose_14")];
+            tensor<fp16, [1, ?, 512]> var_247_cast_fp16 = layer_norm(axes = var_247_axes_0, beta = convnext_3_norm_norm_bias_to_fp16, epsilon = var_218_to_fp16, gamma = convnext_3_norm_norm_weight_to_fp16, x = input_39_cast_fp16)[name = string("op_247_cast_fp16")];
+            tensor<int32, [3]> input_41_perm_0 = const()[name = string("input_41_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            string h_13_pad_type_0 = const()[name = string("h_13_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_13_strides_0 = const()[name = string("h_13_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_13_pad_0 = const()[name = string("h_13_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_13_dilations_0 = const()[name = string("h_13_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_13_groups_0 = const()[name = string("h_13_groups_0"), val = int32(1)];
+            tensor<fp16, [2048, 512, 1]> convnext_3_pwconv1_weight_to_fp16 = const()[name = string("convnext_3_pwconv1_weight_to_fp16"), val = tensor<fp16, [2048, 512, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(12814528)))];
+            tensor<fp16, [2048]> convnext_3_pwconv1_bias_to_fp16 = const()[name = string("convnext_3_pwconv1_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(14911744)))];
+            tensor<fp16, [1, 512, ?]> input_41_cast_fp16 = transpose(perm = input_41_perm_0, x = var_247_cast_fp16)[name = string("transpose_13")];
+            tensor<fp16, [1, 2048, ?]> h_13_cast_fp16 = conv(bias = convnext_3_pwconv1_bias_to_fp16, dilations = h_13_dilations_0, groups = h_13_groups_0, pad = h_13_pad_0, pad_type = h_13_pad_type_0, strides = h_13_strides_0, weight = convnext_3_pwconv1_weight_to_fp16, x = input_41_cast_fp16)[name = string("h_13_cast_fp16")];
+            string input_43_mode_0 = const()[name = string("input_43_mode_0"), val = string("EXACT")];
+            tensor<fp16, [1, 2048, ?]> input_43_cast_fp16 = gelu(mode = input_43_mode_0, x = h_13_cast_fp16)[name = string("input_43_cast_fp16")];
+            string h_15_pad_type_0 = const()[name = string("h_15_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_15_strides_0 = const()[name = string("h_15_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_15_pad_0 = const()[name = string("h_15_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_15_dilations_0 = const()[name = string("h_15_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_15_groups_0 = const()[name = string("h_15_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 2048, 1]> var_264_weight_0_to_fp16 = const()[name = string("op_264_weight_0_to_fp16"), val = tensor<fp16, [512, 2048, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(14915904)))];
+            tensor<fp16, [512]> var_264_bias_0_to_fp16 = const()[name = string("op_264_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(17013120)))];
+            tensor<fp16, [1, 512, ?]> var_264_cast_fp16 = conv(bias = var_264_bias_0_to_fp16, dilations = h_15_dilations_0, groups = h_15_groups_0, pad = h_15_pad_0, pad_type = h_15_pad_type_0, strides = h_15_strides_0, weight = var_264_weight_0_to_fp16, x = input_43_cast_fp16)[name = string("op_264_cast_fp16")];
+            tensor<fp16, [1, 512, ?]> input_45_cast_fp16 = add(x = input_35_cast_fp16, y = var_264_cast_fp16)[name = string("input_45_cast_fp16")];
+            tensor<int32, [6]> input_47_pad_0 = const()[name = string("input_47_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 12, 0])];
+            string input_47_mode_0 = const()[name = string("input_47_mode_0"), val = string("replicate")];
+            fp16 const_5_to_fp16 = const()[name = string("const_5_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 512, ?]> input_47_cast_fp16 = pad(constant_val = const_5_to_fp16, mode = input_47_mode_0, pad = input_47_pad_0, x = input_45_cast_fp16)[name = string("input_47_cast_fp16")];
+            string x_17_pad_type_0 = const()[name = string("x_17_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> x_17_dilations_0 = const()[name = string("x_17_dilations_0"), val = tensor<int32, [1]>([2])];
+            int32 x_17_groups_0 = const()[name = string("x_17_groups_0"), val = int32(512)];
+            tensor<int32, [1]> x_17_strides_0 = const()[name = string("x_17_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> x_17_pad_0 = const()[name = string("x_17_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<fp16, [512, 1, 7]> convnext_4_dwconv_net_weight_to_fp16 = const()[name = string("convnext_4_dwconv_net_weight_to_fp16"), val = tensor<fp16, [512, 1, 7]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(17014208)))];
+            tensor<fp16, [512]> convnext_4_dwconv_net_bias_to_fp16 = const()[name = string("convnext_4_dwconv_net_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(17021440)))];
+            tensor<fp16, [1, 512, ?]> x_17_cast_fp16 = conv(bias = convnext_4_dwconv_net_bias_to_fp16, dilations = x_17_dilations_0, groups = x_17_groups_0, pad = x_17_pad_0, pad_type = x_17_pad_type_0, strides = x_17_strides_0, weight = convnext_4_dwconv_net_weight_to_fp16, x = input_47_cast_fp16)[name = string("x_17_cast_fp16")];
+            tensor<int32, [3]> input_49_perm_0 = const()[name = string("input_49_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [1]> var_297_axes_0 = const()[name = string("op_297_axes_0"), val = tensor<int32, [1]>([-1])];
+            tensor<fp16, [512]> convnext_4_norm_norm_weight_to_fp16 = const()[name = string("convnext_4_norm_norm_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(17022528)))];
+            tensor<fp16, [512]> convnext_4_norm_norm_bias_to_fp16 = const()[name = string("convnext_4_norm_norm_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(17023616)))];
+            fp16 var_267_to_fp16 = const()[name = string("op_267_to_fp16"), val = fp16(0x1.5p-17)];
+            tensor<fp16, [1, ?, 512]> input_49_cast_fp16 = transpose(perm = input_49_perm_0, x = x_17_cast_fp16)[name = string("transpose_12")];
+            tensor<fp16, [1, ?, 512]> var_297_cast_fp16 = layer_norm(axes = var_297_axes_0, beta = convnext_4_norm_norm_bias_to_fp16, epsilon = var_267_to_fp16, gamma = convnext_4_norm_norm_weight_to_fp16, x = input_49_cast_fp16)[name = string("op_297_cast_fp16")];
+            tensor<int32, [3]> input_51_perm_0 = const()[name = string("input_51_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            string h_17_pad_type_0 = const()[name = string("h_17_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_17_strides_0 = const()[name = string("h_17_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_17_pad_0 = const()[name = string("h_17_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_17_dilations_0 = const()[name = string("h_17_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_17_groups_0 = const()[name = string("h_17_groups_0"), val = int32(1)];
+            tensor<fp16, [2048, 512, 1]> convnext_4_pwconv1_weight_to_fp16 = const()[name = string("convnext_4_pwconv1_weight_to_fp16"), val = tensor<fp16, [2048, 512, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(17024704)))];
+            tensor<fp16, [2048]> convnext_4_pwconv1_bias_to_fp16 = const()[name = string("convnext_4_pwconv1_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(19121920)))];
+            tensor<fp16, [1, 512, ?]> input_51_cast_fp16 = transpose(perm = input_51_perm_0, x = var_297_cast_fp16)[name = string("transpose_11")];
+            tensor<fp16, [1, 2048, ?]> h_17_cast_fp16 = conv(bias = convnext_4_pwconv1_bias_to_fp16, dilations = h_17_dilations_0, groups = h_17_groups_0, pad = h_17_pad_0, pad_type = h_17_pad_type_0, strides = h_17_strides_0, weight = convnext_4_pwconv1_weight_to_fp16, x = input_51_cast_fp16)[name = string("h_17_cast_fp16")];
+            string input_53_mode_0 = const()[name = string("input_53_mode_0"), val = string("EXACT")];
+            tensor<fp16, [1, 2048, ?]> input_53_cast_fp16 = gelu(mode = input_53_mode_0, x = h_17_cast_fp16)[name = string("input_53_cast_fp16")];
+            string h_19_pad_type_0 = const()[name = string("h_19_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_19_strides_0 = const()[name = string("h_19_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_19_pad_0 = const()[name = string("h_19_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_19_dilations_0 = const()[name = string("h_19_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_19_groups_0 = const()[name = string("h_19_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 2048, 1]> var_314_weight_0_to_fp16 = const()[name = string("op_314_weight_0_to_fp16"), val = tensor<fp16, [512, 2048, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(19126080)))];
+            tensor<fp16, [512]> var_314_bias_0_to_fp16 = const()[name = string("op_314_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(21223296)))];
+            tensor<fp16, [1, 512, ?]> var_314_cast_fp16 = conv(bias = var_314_bias_0_to_fp16, dilations = h_19_dilations_0, groups = h_19_groups_0, pad = h_19_pad_0, pad_type = h_19_pad_type_0, strides = h_19_strides_0, weight = var_314_weight_0_to_fp16, x = input_53_cast_fp16)[name = string("op_314_cast_fp16")];
+            tensor<fp16, [1, 512, ?]> input_55_cast_fp16 = add(x = input_45_cast_fp16, y = var_314_cast_fp16)[name = string("input_55_cast_fp16")];
+            tensor<int32, [6]> input_57_pad_0 = const()[name = string("input_57_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 24, 0])];
+            string input_57_mode_0 = const()[name = string("input_57_mode_0"), val = string("replicate")];
+            fp16 const_6_to_fp16 = const()[name = string("const_6_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 512, ?]> input_57_cast_fp16 = pad(constant_val = const_6_to_fp16, mode = input_57_mode_0, pad = input_57_pad_0, x = input_55_cast_fp16)[name = string("input_57_cast_fp16")];
+            string x_19_pad_type_0 = const()[name = string("x_19_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> x_19_dilations_0 = const()[name = string("x_19_dilations_0"), val = tensor<int32, [1]>([4])];
+            int32 x_19_groups_0 = const()[name = string("x_19_groups_0"), val = int32(512)];
+            tensor<int32, [1]> x_19_strides_0 = const()[name = string("x_19_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> x_19_pad_0 = const()[name = string("x_19_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<fp16, [512, 1, 7]> convnext_5_dwconv_net_weight_to_fp16 = const()[name = string("convnext_5_dwconv_net_weight_to_fp16"), val = tensor<fp16, [512, 1, 7]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(21224384)))];
+            tensor<fp16, [512]> convnext_5_dwconv_net_bias_to_fp16 = const()[name = string("convnext_5_dwconv_net_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(21231616)))];
+            tensor<fp16, [1, 512, ?]> x_19_cast_fp16 = conv(bias = convnext_5_dwconv_net_bias_to_fp16, dilations = x_19_dilations_0, groups = x_19_groups_0, pad = x_19_pad_0, pad_type = x_19_pad_type_0, strides = x_19_strides_0, weight = convnext_5_dwconv_net_weight_to_fp16, x = input_57_cast_fp16)[name = string("x_19_cast_fp16")];
+            tensor<int32, [3]> input_59_perm_0 = const()[name = string("input_59_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [1]> var_348_axes_0 = const()[name = string("op_348_axes_0"), val = tensor<int32, [1]>([-1])];
+            tensor<fp16, [512]> convnext_5_norm_norm_weight_to_fp16 = const()[name = string("convnext_5_norm_norm_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(21232704)))];
+            tensor<fp16, [512]> convnext_5_norm_norm_bias_to_fp16 = const()[name = string("convnext_5_norm_norm_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(21233792)))];
+            fp16 var_318_to_fp16 = const()[name = string("op_318_to_fp16"), val = fp16(0x1.5p-17)];
+            tensor<fp16, [1, ?, 512]> input_59_cast_fp16 = transpose(perm = input_59_perm_0, x = x_19_cast_fp16)[name = string("transpose_10")];
+            tensor<fp16, [1, ?, 512]> var_348_cast_fp16 = layer_norm(axes = var_348_axes_0, beta = convnext_5_norm_norm_bias_to_fp16, epsilon = var_318_to_fp16, gamma = convnext_5_norm_norm_weight_to_fp16, x = input_59_cast_fp16)[name = string("op_348_cast_fp16")];
+            tensor<int32, [3]> input_61_perm_0 = const()[name = string("input_61_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            string h_21_pad_type_0 = const()[name = string("h_21_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_21_strides_0 = const()[name = string("h_21_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_21_pad_0 = const()[name = string("h_21_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_21_dilations_0 = const()[name = string("h_21_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_21_groups_0 = const()[name = string("h_21_groups_0"), val = int32(1)];
+            tensor<fp16, [2048, 512, 1]> convnext_5_pwconv1_weight_to_fp16 = const()[name = string("convnext_5_pwconv1_weight_to_fp16"), val = tensor<fp16, [2048, 512, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(21234880)))];
+            tensor<fp16, [2048]> convnext_5_pwconv1_bias_to_fp16 = const()[name = string("convnext_5_pwconv1_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(23332096)))];
+            tensor<fp16, [1, 512, ?]> input_61_cast_fp16 = transpose(perm = input_61_perm_0, x = var_348_cast_fp16)[name = string("transpose_9")];
+            tensor<fp16, [1, 2048, ?]> h_21_cast_fp16 = conv(bias = convnext_5_pwconv1_bias_to_fp16, dilations = h_21_dilations_0, groups = h_21_groups_0, pad = h_21_pad_0, pad_type = h_21_pad_type_0, strides = h_21_strides_0, weight = convnext_5_pwconv1_weight_to_fp16, x = input_61_cast_fp16)[name = string("h_21_cast_fp16")];
+            string input_63_mode_0 = const()[name = string("input_63_mode_0"), val = string("EXACT")];
+            tensor<fp16, [1, 2048, ?]> input_63_cast_fp16 = gelu(mode = input_63_mode_0, x = h_21_cast_fp16)[name = string("input_63_cast_fp16")];
+            string h_23_pad_type_0 = const()[name = string("h_23_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_23_strides_0 = const()[name = string("h_23_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_23_pad_0 = const()[name = string("h_23_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_23_dilations_0 = const()[name = string("h_23_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_23_groups_0 = const()[name = string("h_23_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 2048, 1]> var_365_weight_0_to_fp16 = const()[name = string("op_365_weight_0_to_fp16"), val = tensor<fp16, [512, 2048, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(23336256)))];
+            tensor<fp16, [512]> var_365_bias_0_to_fp16 = const()[name = string("op_365_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(25433472)))];
+            tensor<fp16, [1, 512, ?]> var_365_cast_fp16 = conv(bias = var_365_bias_0_to_fp16, dilations = h_23_dilations_0, groups = h_23_groups_0, pad = h_23_pad_0, pad_type = h_23_pad_type_0, strides = h_23_strides_0, weight = var_365_weight_0_to_fp16, x = input_63_cast_fp16)[name = string("op_365_cast_fp16")];
+            tensor<fp16, [1, 512, ?]> input_65_cast_fp16 = add(x = input_55_cast_fp16, y = var_365_cast_fp16)[name = string("input_65_cast_fp16")];
+            tensor<int32, [6]> input_67_pad_0 = const()[name = string("input_67_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 6, 0])];
+            string input_67_mode_0 = const()[name = string("input_67_mode_0"), val = string("replicate")];
+            fp16 const_7_to_fp16 = const()[name = string("const_7_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 512, ?]> input_67_cast_fp16 = pad(constant_val = const_7_to_fp16, mode = input_67_mode_0, pad = input_67_pad_0, x = input_65_cast_fp16)[name = string("input_67_cast_fp16")];
+            string x_21_pad_type_0 = const()[name = string("x_21_pad_type_0"), val = string("valid")];
+            int32 x_21_groups_0 = const()[name = string("x_21_groups_0"), val = int32(512)];
+            tensor<int32, [1]> x_21_strides_0 = const()[name = string("x_21_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> x_21_pad_0 = const()[name = string("x_21_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> x_21_dilations_0 = const()[name = string("x_21_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<fp16, [512, 1, 7]> convnext_6_dwconv_net_weight_to_fp16 = const()[name = string("convnext_6_dwconv_net_weight_to_fp16"), val = tensor<fp16, [512, 1, 7]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(25434560)))];
+            tensor<fp16, [512]> convnext_6_dwconv_net_bias_to_fp16 = const()[name = string("convnext_6_dwconv_net_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(25441792)))];
+            tensor<fp16, [1, 512, ?]> x_21_cast_fp16 = conv(bias = convnext_6_dwconv_net_bias_to_fp16, dilations = x_21_dilations_0, groups = x_21_groups_0, pad = x_21_pad_0, pad_type = x_21_pad_type_0, strides = x_21_strides_0, weight = convnext_6_dwconv_net_weight_to_fp16, x = input_67_cast_fp16)[name = string("x_21_cast_fp16")];
+            tensor<int32, [3]> input_69_perm_0 = const()[name = string("input_69_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [1]> var_398_axes_0 = const()[name = string("op_398_axes_0"), val = tensor<int32, [1]>([-1])];
+            tensor<fp16, [512]> convnext_6_norm_norm_weight_to_fp16 = const()[name = string("convnext_6_norm_norm_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(25442880)))];
+            tensor<fp16, [512]> convnext_6_norm_norm_bias_to_fp16 = const()[name = string("convnext_6_norm_norm_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(25443968)))];
+            fp16 var_369_to_fp16 = const()[name = string("op_369_to_fp16"), val = fp16(0x1.5p-17)];
+            tensor<fp16, [1, ?, 512]> input_69_cast_fp16 = transpose(perm = input_69_perm_0, x = x_21_cast_fp16)[name = string("transpose_8")];
+            tensor<fp16, [1, ?, 512]> var_398_cast_fp16 = layer_norm(axes = var_398_axes_0, beta = convnext_6_norm_norm_bias_to_fp16, epsilon = var_369_to_fp16, gamma = convnext_6_norm_norm_weight_to_fp16, x = input_69_cast_fp16)[name = string("op_398_cast_fp16")];
+            tensor<int32, [3]> input_71_perm_0 = const()[name = string("input_71_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            string h_25_pad_type_0 = const()[name = string("h_25_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_25_strides_0 = const()[name = string("h_25_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_25_pad_0 = const()[name = string("h_25_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_25_dilations_0 = const()[name = string("h_25_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_25_groups_0 = const()[name = string("h_25_groups_0"), val = int32(1)];
+            tensor<fp16, [2048, 512, 1]> convnext_6_pwconv1_weight_to_fp16 = const()[name = string("convnext_6_pwconv1_weight_to_fp16"), val = tensor<fp16, [2048, 512, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(25445056)))];
+            tensor<fp16, [2048]> convnext_6_pwconv1_bias_to_fp16 = const()[name = string("convnext_6_pwconv1_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(27542272)))];
+            tensor<fp16, [1, 512, ?]> input_71_cast_fp16 = transpose(perm = input_71_perm_0, x = var_398_cast_fp16)[name = string("transpose_7")];
+            tensor<fp16, [1, 2048, ?]> h_25_cast_fp16 = conv(bias = convnext_6_pwconv1_bias_to_fp16, dilations = h_25_dilations_0, groups = h_25_groups_0, pad = h_25_pad_0, pad_type = h_25_pad_type_0, strides = h_25_strides_0, weight = convnext_6_pwconv1_weight_to_fp16, x = input_71_cast_fp16)[name = string("h_25_cast_fp16")];
+            string input_73_mode_0 = const()[name = string("input_73_mode_0"), val = string("EXACT")];
+            tensor<fp16, [1, 2048, ?]> input_73_cast_fp16 = gelu(mode = input_73_mode_0, x = h_25_cast_fp16)[name = string("input_73_cast_fp16")];
+            string h_27_pad_type_0 = const()[name = string("h_27_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_27_strides_0 = const()[name = string("h_27_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_27_pad_0 = const()[name = string("h_27_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_27_dilations_0 = const()[name = string("h_27_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_27_groups_0 = const()[name = string("h_27_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 2048, 1]> var_415_weight_0_to_fp16 = const()[name = string("op_415_weight_0_to_fp16"), val = tensor<fp16, [512, 2048, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(27546432)))];
+            tensor<fp16, [512]> var_415_bias_0_to_fp16 = const()[name = string("op_415_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(29643648)))];
+            tensor<fp16, [1, 512, ?]> var_415_cast_fp16 = conv(bias = var_415_bias_0_to_fp16, dilations = h_27_dilations_0, groups = h_27_groups_0, pad = h_27_pad_0, pad_type = h_27_pad_type_0, strides = h_27_strides_0, weight = var_415_weight_0_to_fp16, x = input_73_cast_fp16)[name = string("op_415_cast_fp16")];
+            tensor<fp16, [1, 512, ?]> input_75_cast_fp16 = add(x = input_65_cast_fp16, y = var_415_cast_fp16)[name = string("input_75_cast_fp16")];
+            tensor<int32, [6]> input_77_pad_0 = const()[name = string("input_77_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 6, 0])];
+            string input_77_mode_0 = const()[name = string("input_77_mode_0"), val = string("replicate")];
+            fp16 const_8_to_fp16 = const()[name = string("const_8_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 512, ?]> input_77_cast_fp16 = pad(constant_val = const_8_to_fp16, mode = input_77_mode_0, pad = input_77_pad_0, x = input_75_cast_fp16)[name = string("input_77_cast_fp16")];
+            string x_23_pad_type_0 = const()[name = string("x_23_pad_type_0"), val = string("valid")];
+            int32 x_23_groups_0 = const()[name = string("x_23_groups_0"), val = int32(512)];
+            tensor<int32, [1]> x_23_strides_0 = const()[name = string("x_23_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> x_23_pad_0 = const()[name = string("x_23_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> x_23_dilations_0 = const()[name = string("x_23_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<fp16, [512, 1, 7]> convnext_7_dwconv_net_weight_to_fp16 = const()[name = string("convnext_7_dwconv_net_weight_to_fp16"), val = tensor<fp16, [512, 1, 7]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(29644736)))];
+            tensor<fp16, [512]> convnext_7_dwconv_net_bias_to_fp16 = const()[name = string("convnext_7_dwconv_net_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(29651968)))];
+            tensor<fp16, [1, 512, ?]> x_23_cast_fp16 = conv(bias = convnext_7_dwconv_net_bias_to_fp16, dilations = x_23_dilations_0, groups = x_23_groups_0, pad = x_23_pad_0, pad_type = x_23_pad_type_0, strides = x_23_strides_0, weight = convnext_7_dwconv_net_weight_to_fp16, x = input_77_cast_fp16)[name = string("x_23_cast_fp16")];
+            tensor<int32, [3]> input_79_perm_0 = const()[name = string("input_79_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [1]> var_448_axes_0 = const()[name = string("op_448_axes_0"), val = tensor<int32, [1]>([-1])];
+            tensor<fp16, [512]> convnext_7_norm_norm_weight_to_fp16 = const()[name = string("convnext_7_norm_norm_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(29653056)))];
+            tensor<fp16, [512]> convnext_7_norm_norm_bias_to_fp16 = const()[name = string("convnext_7_norm_norm_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(29654144)))];
+            fp16 var_419_to_fp16 = const()[name = string("op_419_to_fp16"), val = fp16(0x1.5p-17)];
+            tensor<fp16, [1, ?, 512]> input_79_cast_fp16 = transpose(perm = input_79_perm_0, x = x_23_cast_fp16)[name = string("transpose_6")];
+            tensor<fp16, [1, ?, 512]> var_448_cast_fp16 = layer_norm(axes = var_448_axes_0, beta = convnext_7_norm_norm_bias_to_fp16, epsilon = var_419_to_fp16, gamma = convnext_7_norm_norm_weight_to_fp16, x = input_79_cast_fp16)[name = string("op_448_cast_fp16")];
+            tensor<int32, [3]> input_81_perm_0 = const()[name = string("input_81_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            string h_29_pad_type_0 = const()[name = string("h_29_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_29_strides_0 = const()[name = string("h_29_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_29_pad_0 = const()[name = string("h_29_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_29_dilations_0 = const()[name = string("h_29_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_29_groups_0 = const()[name = string("h_29_groups_0"), val = int32(1)];
+            tensor<fp16, [2048, 512, 1]> convnext_7_pwconv1_weight_to_fp16 = const()[name = string("convnext_7_pwconv1_weight_to_fp16"), val = tensor<fp16, [2048, 512, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(29655232)))];
+            tensor<fp16, [2048]> convnext_7_pwconv1_bias_to_fp16 = const()[name = string("convnext_7_pwconv1_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(31752448)))];
+            tensor<fp16, [1, 512, ?]> input_81_cast_fp16 = transpose(perm = input_81_perm_0, x = var_448_cast_fp16)[name = string("transpose_5")];
+            tensor<fp16, [1, 2048, ?]> h_29_cast_fp16 = conv(bias = convnext_7_pwconv1_bias_to_fp16, dilations = h_29_dilations_0, groups = h_29_groups_0, pad = h_29_pad_0, pad_type = h_29_pad_type_0, strides = h_29_strides_0, weight = convnext_7_pwconv1_weight_to_fp16, x = input_81_cast_fp16)[name = string("h_29_cast_fp16")];
+            string input_83_mode_0 = const()[name = string("input_83_mode_0"), val = string("EXACT")];
+            tensor<fp16, [1, 2048, ?]> input_83_cast_fp16 = gelu(mode = input_83_mode_0, x = h_29_cast_fp16)[name = string("input_83_cast_fp16")];
+            string h_31_pad_type_0 = const()[name = string("h_31_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_31_strides_0 = const()[name = string("h_31_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_31_pad_0 = const()[name = string("h_31_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_31_dilations_0 = const()[name = string("h_31_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_31_groups_0 = const()[name = string("h_31_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 2048, 1]> var_465_weight_0_to_fp16 = const()[name = string("op_465_weight_0_to_fp16"), val = tensor<fp16, [512, 2048, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(31756608)))];
+            tensor<fp16, [512]> var_465_bias_0_to_fp16 = const()[name = string("op_465_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(33853824)))];
+            tensor<fp16, [1, 512, ?]> var_465_cast_fp16 = conv(bias = var_465_bias_0_to_fp16, dilations = h_31_dilations_0, groups = h_31_groups_0, pad = h_31_pad_0, pad_type = h_31_pad_type_0, strides = h_31_strides_0, weight = var_465_weight_0_to_fp16, x = input_83_cast_fp16)[name = string("op_465_cast_fp16")];
+            tensor<fp16, [1, 512, ?]> input_85_cast_fp16 = add(x = input_75_cast_fp16, y = var_465_cast_fp16)[name = string("input_85_cast_fp16")];
+            tensor<int32, [6]> input_87_pad_0 = const()[name = string("input_87_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 6, 0])];
+            string input_87_mode_0 = const()[name = string("input_87_mode_0"), val = string("replicate")];
+            fp16 const_9_to_fp16 = const()[name = string("const_9_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 512, ?]> input_87_cast_fp16 = pad(constant_val = const_9_to_fp16, mode = input_87_mode_0, pad = input_87_pad_0, x = input_85_cast_fp16)[name = string("input_87_cast_fp16")];
+            string x_25_pad_type_0 = const()[name = string("x_25_pad_type_0"), val = string("valid")];
+            int32 x_25_groups_0 = const()[name = string("x_25_groups_0"), val = int32(512)];
+            tensor<int32, [1]> x_25_strides_0 = const()[name = string("x_25_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> x_25_pad_0 = const()[name = string("x_25_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> x_25_dilations_0 = const()[name = string("x_25_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<fp16, [512, 1, 7]> convnext_8_dwconv_net_weight_to_fp16 = const()[name = string("convnext_8_dwconv_net_weight_to_fp16"), val = tensor<fp16, [512, 1, 7]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(33854912)))];
+            tensor<fp16, [512]> convnext_8_dwconv_net_bias_to_fp16 = const()[name = string("convnext_8_dwconv_net_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(33862144)))];
+            tensor<fp16, [1, 512, ?]> x_25_cast_fp16 = conv(bias = convnext_8_dwconv_net_bias_to_fp16, dilations = x_25_dilations_0, groups = x_25_groups_0, pad = x_25_pad_0, pad_type = x_25_pad_type_0, strides = x_25_strides_0, weight = convnext_8_dwconv_net_weight_to_fp16, x = input_87_cast_fp16)[name = string("x_25_cast_fp16")];
+            tensor<int32, [3]> input_89_perm_0 = const()[name = string("input_89_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [1]> var_498_axes_0 = const()[name = string("op_498_axes_0"), val = tensor<int32, [1]>([-1])];
+            tensor<fp16, [512]> convnext_8_norm_norm_weight_to_fp16 = const()[name = string("convnext_8_norm_norm_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(33863232)))];
+            tensor<fp16, [512]> convnext_8_norm_norm_bias_to_fp16 = const()[name = string("convnext_8_norm_norm_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(33864320)))];
+            fp16 var_469_to_fp16 = const()[name = string("op_469_to_fp16"), val = fp16(0x1.5p-17)];
+            tensor<fp16, [1, ?, 512]> input_89_cast_fp16 = transpose(perm = input_89_perm_0, x = x_25_cast_fp16)[name = string("transpose_4")];
+            tensor<fp16, [1, ?, 512]> var_498_cast_fp16 = layer_norm(axes = var_498_axes_0, beta = convnext_8_norm_norm_bias_to_fp16, epsilon = var_469_to_fp16, gamma = convnext_8_norm_norm_weight_to_fp16, x = input_89_cast_fp16)[name = string("op_498_cast_fp16")];
+            tensor<int32, [3]> input_91_perm_0 = const()[name = string("input_91_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            string h_33_pad_type_0 = const()[name = string("h_33_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_33_strides_0 = const()[name = string("h_33_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_33_pad_0 = const()[name = string("h_33_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_33_dilations_0 = const()[name = string("h_33_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_33_groups_0 = const()[name = string("h_33_groups_0"), val = int32(1)];
+            tensor<fp16, [2048, 512, 1]> convnext_8_pwconv1_weight_to_fp16 = const()[name = string("convnext_8_pwconv1_weight_to_fp16"), val = tensor<fp16, [2048, 512, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(33865408)))];
+            tensor<fp16, [2048]> convnext_8_pwconv1_bias_to_fp16 = const()[name = string("convnext_8_pwconv1_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(35962624)))];
+            tensor<fp16, [1, 512, ?]> input_91_cast_fp16 = transpose(perm = input_91_perm_0, x = var_498_cast_fp16)[name = string("transpose_3")];
+            tensor<fp16, [1, 2048, ?]> h_33_cast_fp16 = conv(bias = convnext_8_pwconv1_bias_to_fp16, dilations = h_33_dilations_0, groups = h_33_groups_0, pad = h_33_pad_0, pad_type = h_33_pad_type_0, strides = h_33_strides_0, weight = convnext_8_pwconv1_weight_to_fp16, x = input_91_cast_fp16)[name = string("h_33_cast_fp16")];
+            string input_93_mode_0 = const()[name = string("input_93_mode_0"), val = string("EXACT")];
+            tensor<fp16, [1, 2048, ?]> input_93_cast_fp16 = gelu(mode = input_93_mode_0, x = h_33_cast_fp16)[name = string("input_93_cast_fp16")];
+            string h_35_pad_type_0 = const()[name = string("h_35_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_35_strides_0 = const()[name = string("h_35_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_35_pad_0 = const()[name = string("h_35_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_35_dilations_0 = const()[name = string("h_35_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_35_groups_0 = const()[name = string("h_35_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 2048, 1]> var_515_weight_0_to_fp16 = const()[name = string("op_515_weight_0_to_fp16"), val = tensor<fp16, [512, 2048, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(35966784)))];
+            tensor<fp16, [512]> var_515_bias_0_to_fp16 = const()[name = string("op_515_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(38064000)))];
+            tensor<fp16, [1, 512, ?]> var_515_cast_fp16 = conv(bias = var_515_bias_0_to_fp16, dilations = h_35_dilations_0, groups = h_35_groups_0, pad = h_35_pad_0, pad_type = h_35_pad_type_0, strides = h_35_strides_0, weight = var_515_weight_0_to_fp16, x = input_93_cast_fp16)[name = string("op_515_cast_fp16")];
+            tensor<fp16, [1, 512, ?]> input_95_cast_fp16 = add(x = input_85_cast_fp16, y = var_515_cast_fp16)[name = string("input_95_cast_fp16")];
+            tensor<int32, [6]> input_97_pad_0 = const()[name = string("input_97_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 6, 0])];
+            string input_97_mode_0 = const()[name = string("input_97_mode_0"), val = string("replicate")];
+            fp16 const_10_to_fp16 = const()[name = string("const_10_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 512, ?]> input_97_cast_fp16 = pad(constant_val = const_10_to_fp16, mode = input_97_mode_0, pad = input_97_pad_0, x = input_95_cast_fp16)[name = string("input_97_cast_fp16")];
+            string x_27_pad_type_0 = const()[name = string("x_27_pad_type_0"), val = string("valid")];
+            int32 x_27_groups_0 = const()[name = string("x_27_groups_0"), val = int32(512)];
+            tensor<int32, [1]> x_27_strides_0 = const()[name = string("x_27_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> x_27_pad_0 = const()[name = string("x_27_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> x_27_dilations_0 = const()[name = string("x_27_dilations_0"), val = tensor<int32, [1]>([1])];
+            tensor<fp16, [512, 1, 7]> convnext_9_dwconv_net_weight_to_fp16 = const()[name = string("convnext_9_dwconv_net_weight_to_fp16"), val = tensor<fp16, [512, 1, 7]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(38065088)))];
+            tensor<fp16, [512]> convnext_9_dwconv_net_bias_to_fp16 = const()[name = string("convnext_9_dwconv_net_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(38072320)))];
+            tensor<fp16, [1, 512, ?]> x_27_cast_fp16 = conv(bias = convnext_9_dwconv_net_bias_to_fp16, dilations = x_27_dilations_0, groups = x_27_groups_0, pad = x_27_pad_0, pad_type = x_27_pad_type_0, strides = x_27_strides_0, weight = convnext_9_dwconv_net_weight_to_fp16, x = input_97_cast_fp16)[name = string("x_27_cast_fp16")];
+            tensor<int32, [3]> input_99_perm_0 = const()[name = string("input_99_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [1]> var_548_axes_0 = const()[name = string("op_548_axes_0"), val = tensor<int32, [1]>([-1])];
+            tensor<fp16, [512]> convnext_9_norm_norm_weight_to_fp16 = const()[name = string("convnext_9_norm_norm_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(38073408)))];
+            tensor<fp16, [512]> convnext_9_norm_norm_bias_to_fp16 = const()[name = string("convnext_9_norm_norm_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(38074496)))];
+            fp16 var_519_to_fp16 = const()[name = string("op_519_to_fp16"), val = fp16(0x1.5p-17)];
+            tensor<fp16, [1, ?, 512]> input_99_cast_fp16 = transpose(perm = input_99_perm_0, x = x_27_cast_fp16)[name = string("transpose_2")];
+            tensor<fp16, [1, ?, 512]> var_548_cast_fp16 = layer_norm(axes = var_548_axes_0, beta = convnext_9_norm_norm_bias_to_fp16, epsilon = var_519_to_fp16, gamma = convnext_9_norm_norm_weight_to_fp16, x = input_99_cast_fp16)[name = string("op_548_cast_fp16")];
+            tensor<int32, [3]> input_101_perm_0 = const()[name = string("input_101_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            string h_37_pad_type_0 = const()[name = string("h_37_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_37_strides_0 = const()[name = string("h_37_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_37_pad_0 = const()[name = string("h_37_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_37_dilations_0 = const()[name = string("h_37_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_37_groups_0 = const()[name = string("h_37_groups_0"), val = int32(1)];
+            tensor<fp16, [2048, 512, 1]> convnext_9_pwconv1_weight_to_fp16 = const()[name = string("convnext_9_pwconv1_weight_to_fp16"), val = tensor<fp16, [2048, 512, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(38075584)))];
+            tensor<fp16, [2048]> convnext_9_pwconv1_bias_to_fp16 = const()[name = string("convnext_9_pwconv1_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(40172800)))];
+            tensor<fp16, [1, 512, ?]> input_101_cast_fp16 = transpose(perm = input_101_perm_0, x = var_548_cast_fp16)[name = string("transpose_1")];
+            tensor<fp16, [1, 2048, ?]> h_37_cast_fp16 = conv(bias = convnext_9_pwconv1_bias_to_fp16, dilations = h_37_dilations_0, groups = h_37_groups_0, pad = h_37_pad_0, pad_type = h_37_pad_type_0, strides = h_37_strides_0, weight = convnext_9_pwconv1_weight_to_fp16, x = input_101_cast_fp16)[name = string("h_37_cast_fp16")];
+            string input_103_mode_0 = const()[name = string("input_103_mode_0"), val = string("EXACT")];
+            tensor<fp16, [1, 2048, ?]> input_103_cast_fp16 = gelu(mode = input_103_mode_0, x = h_37_cast_fp16)[name = string("input_103_cast_fp16")];
+            string h_pad_type_0 = const()[name = string("h_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> h_strides_0 = const()[name = string("h_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> h_pad_0 = const()[name = string("h_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> h_dilations_0 = const()[name = string("h_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 h_groups_0 = const()[name = string("h_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 2048, 1]> var_565_weight_0_to_fp16 = const()[name = string("op_565_weight_0_to_fp16"), val = tensor<fp16, [512, 2048, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(40176960)))];
+            tensor<fp16, [512]> var_565_bias_0_to_fp16 = const()[name = string("op_565_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(42274176)))];
+            tensor<fp16, [1, 512, ?]> var_565_cast_fp16 = conv(bias = var_565_bias_0_to_fp16, dilations = h_dilations_0, groups = h_groups_0, pad = h_pad_0, pad_type = h_pad_type_0, strides = h_strides_0, weight = var_565_weight_0_to_fp16, x = input_103_cast_fp16)[name = string("op_565_cast_fp16")];
+            tensor<fp16, [1, 512, ?]> input_105_cast_fp16 = add(x = input_95_cast_fp16, y = var_565_cast_fp16)[name = string("input_105_cast_fp16")];
+            tensor<fp16, [512]> final_norm_norm_running_mean_to_fp16 = const()[name = string("final_norm_norm_running_mean_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(42275264)))];
+            tensor<fp16, [512]> final_norm_norm_running_var_to_fp16 = const()[name = string("final_norm_norm_running_var_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(42276352)))];
+            tensor<fp16, [512]> final_norm_norm_weight_to_fp16 = const()[name = string("final_norm_norm_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(42277440)))];
+            tensor<fp16, [512]> final_norm_norm_bias_to_fp16 = const()[name = string("final_norm_norm_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(42278528)))];
+            fp16 var_569_to_fp16 = const()[name = string("op_569_to_fp16"), val = fp16(0x1.5p-17)];
+            tensor<fp16, [1, 512, ?]> input_107_cast_fp16 = batch_norm(beta = final_norm_norm_bias_to_fp16, epsilon = var_569_to_fp16, gamma = final_norm_norm_weight_to_fp16, mean = final_norm_norm_running_mean_to_fp16, variance = final_norm_norm_running_var_to_fp16, x = input_105_cast_fp16)[name = string("input_107_cast_fp16")];
+            tensor<int32, [6]> input_109_pad_0 = const()[name = string("input_109_pad_0"), val = tensor<int32, [6]>([0, 0, 0, 0, 2, 0])];
+            string input_109_mode_0 = const()[name = string("input_109_mode_0"), val = string("replicate")];
+            fp16 const_11_to_fp16 = const()[name = string("const_11_to_fp16"), val = fp16(0x0p+0)];
+            tensor<fp16, [1, 512, ?]> input_109_cast_fp16 = pad(constant_val = const_11_to_fp16, mode = input_109_mode_0, pad = input_109_pad_0, x = input_107_cast_fp16)[name = string("input_109_cast_fp16")];
+            string input_111_pad_type_0 = const()[name = string("input_111_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> input_111_strides_0 = const()[name = string("input_111_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> input_111_pad_0 = const()[name = string("input_111_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> input_111_dilations_0 = const()[name = string("input_111_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 input_111_groups_0 = const()[name = string("input_111_groups_0"), val = int32(1)];
+            tensor<fp16, [2048, 512, 3]> head_layer1_net_weight_to_fp16 = const()[name = string("head_layer1_net_weight_to_fp16"), val = tensor<fp16, [2048, 512, 3]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(42279616)))];
+            tensor<fp16, [2048]> head_layer1_net_bias_to_fp16 = const()[name = string("head_layer1_net_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(48571136)))];
+            tensor<fp16, [1, 2048, ?]> input_111_cast_fp16 = conv(bias = head_layer1_net_bias_to_fp16, dilations = input_111_dilations_0, groups = input_111_groups_0, pad = input_111_pad_0, pad_type = input_111_pad_type_0, strides = input_111_strides_0, weight = head_layer1_net_weight_to_fp16, x = input_109_cast_fp16)[name = string("input_111_cast_fp16")];
+            fp32 input_alpha_1 = const()[name = string("input_alpha_1"), val = fp32(0x1.9814fap-13)];
+            tensor<fp16, [1, 2048, ?]> input_cast_fp16 = leaky_relu(alpha = input_alpha_1, x = input_111_cast_fp16)[name = string("input_cast_fp16")];
+            string x_29_pad_type_0 = const()[name = string("x_29_pad_type_0"), val = string("valid")];
+            tensor<int32, [1]> x_29_strides_0 = const()[name = string("x_29_strides_0"), val = tensor<int32, [1]>([1])];
+            tensor<int32, [2]> x_29_pad_0 = const()[name = string("x_29_pad_0"), val = tensor<int32, [2]>([0, 0])];
+            tensor<int32, [1]> x_29_dilations_0 = const()[name = string("x_29_dilations_0"), val = tensor<int32, [1]>([1])];
+            int32 x_29_groups_0 = const()[name = string("x_29_groups_0"), val = int32(1)];
+            tensor<fp16, [512, 2048, 1]> head_layer2_weight_to_fp16 = const()[name = string("head_layer2_weight_to_fp16"), val = tensor<fp16, [512, 2048, 1]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(48575296)))];
+            tensor<fp16, [1, 512, ?]> x_29_cast_fp16 = conv(dilations = x_29_dilations_0, groups = x_29_groups_0, pad = x_29_pad_0, pad_type = x_29_pad_type_0, strides = x_29_strides_0, weight = head_layer2_weight_to_fp16, x = input_cast_fp16)[name = string("x_29_cast_fp16")];
+            tensor<int32, [3]> var_607_perm_0 = const()[name = string("op_607_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [2]> var_612 = const()[name = string("op_612"), val = tensor<int32, [2]>([1, -1])];
+            tensor<fp16, [1, ?, 512]> var_607_cast_fp16 = transpose(perm = var_607_perm_0, x = x_29_cast_fp16)[name = string("transpose_0")];
+            tensor<fp16, [1, ?]> var_613_cast_fp16 = reshape(shape = var_612, x = var_607_cast_fp16)[name = string("op_613_cast_fp16")];
+            string var_613_cast_fp16_to_fp32_dtype_0 = const()[name = string("op_613_cast_fp16_to_fp32_dtype_0"), val = string("fp32")];
+            tensor<fp32, [1, ?]> wav = cast(dtype = var_613_cast_fp16_to_fp32_dtype_0, x = var_613_cast_fp16)[name = string("cast_5")];
+        } -> (wav);
+}

Vocoder.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b45e4c17a45cc4de30dcedae9f662548aa35f8f9763632c342b8bbef3089fcee
+size 50672512

files.txt

@@ -0,0 +1,28 @@
+TextEncoder_v2.mlmodelc/analytics/coremldata.bin 243 973b90afda5d837c037cef2dcbbda275f005eeeb55f2712ca6fbcfa68220f5b5
+TextEncoder_v2.mlmodelc/coremldata.bin 432 c1177e3e03cc4130e5dca18eb007f20573bf95d3938abf62ab2d3470b04cbe7f
+TextEncoder_v2.mlmodelc/model.mil 158547 b0fd5d9a34a301221194a90aa2ba9133b34ed981209f5e6c7b89733c1491074b
+TextEncoder_v2.mlmodelc/weights/weight.bin 18022016 7c7afa5d02426a8a363c9cef0e27ce64a5c57a67d1a59f60aac0ac58ffb9877f
+DurationPredictor_v2.mlmodelc/analytics/coremldata.bin 243 d8bfa7996e1b4c0e26c9c1edc3861f28726bbd30f74d4a7519e69ef78c1b2d1c
+DurationPredictor_v2.mlmodelc/coremldata.bin 426 f4f1630670632dd149787fb6ea4f934ab915c390a88d87a4e825ca58f041b4b2
+DurationPredictor_v2.mlmodelc/model.mil 105229 0c9c67236d575b2b8f0077815e0b57a30cc7d770dead3ad068fc060c68e33920
+DurationPredictor_v2.mlmodelc/weights/weight.bin 1797952 7f709c544087966d6b2538ac50b7a4014d6e1049e21d713cf3bc71c4c5b9307c
+VectorEstimatorVariants/VectorEstimator_v2_int8.mlmodelc/analytics/coremldata.bin 243 d27fc7601864cc7e4ee95a4d33f7202a014fcf548fd89f3a50fe1cf83900c1e5
+VectorEstimatorVariants/VectorEstimator_v2_int8.mlmodelc/coremldata.bin 634 8a61ba6647e635b19ba1bbc39d4e3fa35ad5c18bef4f9513c3137a2a0d4a6705
+VectorEstimatorVariants/VectorEstimator_v2_int8.mlmodelc/model.mil 395983 9c6e37dbaa5ac80377fed83234b228a75797313849392f2fe74560f3575b8dbc
+VectorEstimatorVariants/VectorEstimator_v2_int8.mlmodelc/weights/weight.bin 64184608 0e637c3a1725b5e1b3df09cc162d1822e8fe80fa92ba01c180dde2ca29c12523
+Vocoder.mlmodelc/analytics/coremldata.bin 243 4fe1b825137629a96dc58a1339bc4ece32041b755f99d638f21a153f2e7faed6
+Vocoder.mlmodelc/coremldata.bin 388 dbf54fc98879ee1c9a213452e16d09367f3c34ad489b175bcae7ef5f74f8d121
+Vocoder.mlmodelc/model.mil 75626 5f57d6b16a4ee7237e046a423c265aa2c487857eb0003bba1c89fa66e577353f
+Vocoder.mlmodelc/weights/weight.bin 50672512 b45e4c17a45cc4de30dcedae9f662548aa35f8f9763632c342b8bbef3089fcee
+voice_styles/F1.json 292046 bbdec6ee00231c2c742ad05483df5334cab3b52fda3ba38e6a07059c4563dbc2
+voice_styles/F2.json 292423 7c722c6a72707b1a77f035d67f0d1351ba187738e06f7683e8c72b1df3477fc6
+voice_styles/F3.json 290794 12f6ef2573baa2defa1128069cb59f203e3ab67c92af77b42df8a0e3a2f7c6ab
+voice_styles/F4.json 291808 c2fa764c1225a76dfc3e2c73e8aa4f70d9ee48793860eb34c295fff01c2e032b
+voice_styles/F5.json 291479 45966e73316415626cf41a7d1c6f3b4c70dbc1ba2bee5c1978ef0ce33244fc8d
+voice_styles/M1.json 291748 e35604687f5d23694b8e91593a93eec0e4eca6c0b02bb8ed69139ab2ea6b0a5b
+voice_styles/M2.json 292055 b76cbf62bac707c710cf0ae5aba5e31eea1a6339a9734bfae33ab98499534a50
+voice_styles/M3.json 290198 ea1ac35ccb91b0d7ecad533a2fbd0eec10c91513d8951e3b25fbba99954e159b
+voice_styles/M4.json 291522 ca8eefad4fcd989c9379032ff3e50738adc547eeb5e221b82593a6d7b3bac303
+voice_styles/M5.json 291469 dd22b92740314321f8ae11c5e87f8dd60d060f15dd3a632b5adf77f471f77af2
+tts.json 8253 42078d3aef1cd43ab43021f3c54f47d2d75ceb4e75f627f118890128b06a0d09
+unicode_indexer.json 277676 9bf7346e43883a81f8645c81224f786d43c5b57f3641f6e7671a7d6c493cb24f

tts.json

@@ -0,0 +1,311 @@
+{
+    "tts_version": "v1.7.3",
+    "split": "opensource-multilingual",
+    "ttl": {
+        "latent_dim": 24,
+        "chunk_compress_factor": 6,
+        "batch_expander": {
+            "n_batch_expand": 6
+        },
+        "normalizer": {
+            "scale": 0.25
+        },
+        "text_encoder": {
+            "n_langs": 0,
+            "lang_emb_dim": 0,
+            "text_embedder": {
+                "char_emb_dim": 256
+            },
+            "convnext": {
+                "idim": 256,
+                "ksz": 5,
+                "intermediate_dim": 1024,
+                "num_layers": 6,
+                "dilation_lst": [
+                    1,
+                    1,
+                    2,
+                    2,
+                    4,
+                    4
+                ]
+            },
+            "attn_encoder": {
+                "hidden_channels": 256,
+                "filter_channels": 1024,
+                "n_heads": 4,
+                "n_layers": 4,
+                "p_dropout": 0.0
+            },
+            "proj_out": {
+                "idim": 256,
+                "odim": 256
+            }
+        },
+        "flow_matching": {
+            "sig_min": 1e-08
+        },
+        "style_encoder": {
+            "proj_in": {
+                "ldim": 24,
+                "chunk_compress_factor": 6,
+                "odim": 256
+            },
+            "convnext": {
+                "idim": 256,
+                "ksz": 5,
+                "intermediate_dim": 1024,
+                "num_layers": 6,
+                "dilation_lst": [
+                    1,
+                    1,
+                    1,
+                    1,
+                    1,
+                    1
+                ]
+            },
+            "style_token_layer": {
+                "input_dim": 256,
+                "n_style": 50,
+                "style_key_dim": 256,
+                "style_value_dim": 256,
+                "prototype_dim": 256,
+                "n_units": 256,
+                "n_heads": 2
+            }
+        },
+        "speech_prompted_text_encoder": {
+            "text_dim": 256,
+            "style_dim": 256,
+            "n_units": 256,
+            "n_heads": 2
+        },
+        "uncond_masker": {
+            "prob_both_uncond": 0.04,
+            "prob_text_uncond": 0.01,
+            "std": 0.1,
+            "text_dim": 256,
+            "n_style": 50,
+            "style_key_dim": 256,
+            "style_value_dim": 256
+        },
+        "vector_field": {
+            "n_langs": 0,
+            "lang_emb_dim": 0,
+            "proj_in": {
+                "ldim": 24,
+                "chunk_compress_factor": 6,
+                "odim": 512
+            },
+            "time_encoder": {
+                "time_dim": 64,
+                "hdim": 256
+            },
+            "main_blocks": {
+                "n_blocks": 4,
+                "time_cond_layer": {
+                    "idim": 512,
+                    "time_dim": 64
+                },
+                "style_cond_layer": {
+                    "idim": 512,
+                    "style_dim": 256
+                },
+                "text_cond_layer": {
+                    "idim": 512,
+                    "text_dim": 256,
+                    "n_heads": 8,
+                    "n_units": 512,
+                    "use_residual": true,
+                    "rotary_base": 10000,
+                    "rotary_scale": 10
+                },
+                "convnext_0": {
+                    "idim": 512,
+                    "ksz": 5,
+                    "intermediate_dim": 2048,
+                    "num_layers": 4,
+                    "dilation_lst": [
+                        1,
+                        2,
+                        4,
+                        8
+                    ]
+                },
+                "convnext_1": {
+                    "idim": 512,
+                    "ksz": 5,
+                    "intermediate_dim": 2048,
+                    "num_layers": 1,
+                    "dilation_lst": [
+                        1
+                    ]
+                },
+                "convnext_2": {
+                    "idim": 512,
+                    "ksz": 5,
+                    "intermediate_dim": 2048,
+                    "num_layers": 1,
+                    "dilation_lst": [
+                        1
+                    ]
+                }
+            },
+            "last_convnext": {
+                "idim": 512,
+                "ksz": 5,
+                "intermediate_dim": 2048,
+                "num_layers": 4,
+                "dilation_lst": [
+                    1,
+                    1,
+                    1,
+                    1
+                ]
+            },
+            "proj_out": {
+                "idim": 512,
+                "chunk_compress_factor": 6,
+                "ldim": 24
+            }
+        }
+    },
+    "ae": {
+        "sample_rate": 44100,
+        "n_delay": 0,
+        "base_chunk_size": 512,
+        "chunk_compress_factor": 1,
+        "ldim": 24,
+        "encoder": {
+            "spec_processor": {
+                "n_fft": 2048,
+                "win_length": 2048,
+                "hop_length": 512,
+                "n_mels": 228,
+                "sample_rate": 44100,
+                "eps": 1e-05,
+                "norm_mean": 0.0,
+                "norm_std": 1.0
+            },
+            "ksz_init": 7,
+            "ksz": 7,
+            "num_layers": 10,
+            "dilation_lst": [
+                1,
+                1,
+                1,
+                1,
+                1,
+                1,
+                1,
+                1,
+                1,
+                1
+            ],
+            "intermediate_dim": 2048,
+            "idim": 1253,
+            "hdim": 512,
+            "odim": 24
+        },
+        "decoder": {
+            "ksz_init": 7,
+            "ksz": 7,
+            "num_layers": 10,
+            "dilation_lst": [
+                1,
+                2,
+                4,
+                1,
+                2,
+                4,
+                1,
+                1,
+                1,
+                1
+            ],
+            "intermediate_dim": 2048,
+            "idim": 24,
+            "hdim": 512,
+            "head": {
+                "idim": 512,
+                "hdim": 2048,
+                "odim": 512,
+                "ksz": 3
+            }
+        }
+    },
+    "dp": {
+        "latent_dim": 24,
+        "chunk_compress_factor": 6,
+        "normalizer": {
+            "scale": 1.0
+        },
+        "sentence_encoder": {
+            "char_emb_dim": 64,
+            "text_embedder": {
+                "char_emb_dim": 64
+            },
+            "convnext": {
+                "idim": 64,
+                "ksz": 5,
+                "intermediate_dim": 256,
+                "num_layers": 6,
+                "dilation_lst": [
+                    1,
+                    1,
+                    1,
+                    1,
+                    1,
+                    1
+                ]
+            },
+            "attn_encoder": {
+                "hidden_channels": 64,
+                "filter_channels": 256,
+                "n_heads": 2,
+                "n_layers": 2,
+                "p_dropout": 0.0
+            },
+            "proj_out": {
+                "idim": 64,
+                "odim": 64
+            }
+        },
+        "style_encoder": {
+            "proj_in": {
+                "ldim": 24,
+                "chunk_compress_factor": 6,
+                "odim": 64
+            },
+            "convnext": {
+                "idim": 64,
+                "ksz": 5,
+                "intermediate_dim": 256,
+                "num_layers": 4,
+                "dilation_lst": [
+                    1,
+                    1,
+                    1,
+                    1
+                ]
+            },
+            "style_token_layer": {
+                "input_dim": 64,
+                "n_style": 8,
+                "style_key_dim": 0,
+                "style_value_dim": 16,
+                "prototype_dim": 64,
+                "n_units": 64,
+                "n_heads": 2
+            }
+        },
+        "predictor": {
+            "sentence_dim": 64,
+            "n_style": 8,
+            "style_dim": 16,
+            "hdim": 128,
+            "n_layer": 2
+        }
+    }
+}