altbit commited on
Commit
9b1a9db
·
verified ·
1 Parent(s): bf56dc0

Add files using upload-large-folder tool

Browse files
Files changed (39) hide show
  1. LICENSE +202 -0
  2. README.flux.upstream.md +81 -0
  3. README.md +49 -0
  4. flux/model_index.json +32 -0
  5. flux/scheduler/scheduler_config.json +11 -0
  6. flux/text_encoder/config.json +25 -0
  7. flux/text_encoder/model.safetensors +3 -0
  8. flux/text_encoder_2/config.json +32 -0
  9. flux/text_encoder_2/model-00001-of-00002.safetensors +3 -0
  10. flux/text_encoder_2/model-00002-of-00002.safetensors +3 -0
  11. flux/text_encoder_2/model.safetensors.index.json +226 -0
  12. flux/tokenizer/merges.txt +0 -0
  13. flux/tokenizer/special_tokens_map.json +30 -0
  14. flux/tokenizer/tokenizer_config.json +30 -0
  15. flux/tokenizer/vocab.json +0 -0
  16. flux/tokenizer_2/special_tokens_map.json +125 -0
  17. flux/tokenizer_2/spiece.model +3 -0
  18. flux/tokenizer_2/tokenizer.json +0 -0
  19. flux/tokenizer_2/tokenizer_config.json +940 -0
  20. flux/transformer/config.json +13 -0
  21. flux/transformer/diffusion_pytorch_model-00001-of-00003.safetensors +3 -0
  22. flux/transformer/diffusion_pytorch_model-00002-of-00003.safetensors +3 -0
  23. flux/transformer/diffusion_pytorch_model-00003-of-00003.safetensors +3 -0
  24. flux/transformer/diffusion_pytorch_model.safetensors.index.json +0 -0
  25. flux/vae/config.json +37 -0
  26. flux/vae/diffusion_pytorch_model.safetensors +3 -0
  27. qwen3-tts/.gitattributes +35 -0
  28. qwen3-tts/README.md +1315 -0
  29. qwen3-tts/config.json +167 -0
  30. qwen3-tts/generation_config.json +12 -0
  31. qwen3-tts/merges.txt +0 -0
  32. qwen3-tts/model.safetensors +3 -0
  33. qwen3-tts/preprocessor_config.json +6 -0
  34. qwen3-tts/speech_tokenizer/config.json +94 -0
  35. qwen3-tts/speech_tokenizer/configuration.json +1 -0
  36. qwen3-tts/speech_tokenizer/model.safetensors +3 -0
  37. qwen3-tts/speech_tokenizer/preprocessor_config.json +10 -0
  38. qwen3-tts/tokenizer_config.json +316 -0
  39. qwen3-tts/vocab.json +0 -0
LICENSE ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ Apache License
3
+ Version 2.0, January 2004
4
+ http://www.apache.org/licenses/
5
+
6
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
7
+
8
+ 1. Definitions.
9
+
10
+ "License" shall mean the terms and conditions for use, reproduction,
11
+ and distribution as defined by Sections 1 through 9 of this document.
12
+
13
+ "Licensor" shall mean the copyright owner or entity authorized by
14
+ the copyright owner that is granting the License.
15
+
16
+ "Legal Entity" shall mean the union of the acting entity and all
17
+ other entities that control, are controlled by, or are under common
18
+ control with that entity. For the purposes of this definition,
19
+ "control" means (i) the power, direct or indirect, to cause the
20
+ direction or management of such entity, whether by contract or
21
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
22
+ outstanding shares, or (iii) beneficial ownership of such entity.
23
+
24
+ "You" (or "Your") shall mean an individual or Legal Entity
25
+ exercising permissions granted by this License.
26
+
27
+ "Source" form shall mean the preferred form for making modifications,
28
+ including but not limited to software source code, documentation
29
+ source, and configuration files.
30
+
31
+ "Object" form shall mean any form resulting from mechanical
32
+ transformation or translation of a Source form, including but
33
+ not limited to compiled object code, generated documentation,
34
+ and conversions to other media types.
35
+
36
+ "Work" shall mean the work of authorship, whether in Source or
37
+ Object form, made available under the License, as indicated by a
38
+ copyright notice that is included in or attached to the work
39
+ (an example is provided in the Appendix below).
40
+
41
+ "Derivative Works" shall mean any work, whether in Source or Object
42
+ form, that is based on (or derived from) the Work and for which the
43
+ editorial revisions, annotations, elaborations, or other modifications
44
+ represent, as a whole, an original work of authorship. For the purposes
45
+ of this License, Derivative Works shall not include works that remain
46
+ separable from, or merely link (or bind by name) to the interfaces of,
47
+ the Work and Derivative Works thereof.
48
+
49
+ "Contribution" shall mean any work of authorship, including
50
+ the original version of the Work and any modifications or additions
51
+ to that Work or Derivative Works thereof, that is intentionally
52
+ submitted to Licensor for inclusion in the Work by the copyright owner
53
+ or by an individual or Legal Entity authorized to submit on behalf of
54
+ the copyright owner. For the purposes of this definition, "submitted"
55
+ means any form of electronic, verbal, or written communication sent
56
+ to the Licensor or its representatives, including but not limited to
57
+ communication on electronic mailing lists, source code control systems,
58
+ and issue tracking systems that are managed by, or on behalf of, the
59
+ Licensor for the purpose of discussing and improving the Work, but
60
+ excluding communication that is conspicuously marked or otherwise
61
+ designated in writing by the copyright owner as "Not a Contribution."
62
+
63
+ "Contributor" shall mean Licensor and any individual or Legal Entity
64
+ on behalf of whom a Contribution has been received by Licensor and
65
+ subsequently incorporated within the Work.
66
+
67
+ 2. Grant of Copyright License. Subject to the terms and conditions of
68
+ this License, each Contributor hereby grants to You a perpetual,
69
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
70
+ copyright license to reproduce, prepare Derivative Works of,
71
+ publicly display, publicly perform, sublicense, and distribute the
72
+ Work and such Derivative Works in Source or Object form.
73
+
74
+ 3. Grant of Patent License. Subject to the terms and conditions of
75
+ this License, each Contributor hereby grants to You a perpetual,
76
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
77
+ (except as stated in this section) patent license to make, have made,
78
+ use, offer to sell, sell, import, and otherwise transfer the Work,
79
+ where such license applies only to those patent claims licensable
80
+ by such Contributor that are necessarily infringed by their
81
+ Contribution(s) alone or by combination of their Contribution(s)
82
+ with the Work to which such Contribution(s) was submitted. If You
83
+ institute patent litigation against any entity (including a
84
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
85
+ or a Contribution incorporated within the Work constitutes direct
86
+ or contributory patent infringement, then any patent licenses
87
+ granted to You under this License for that Work shall terminate
88
+ as of the date such litigation is filed.
89
+
90
+ 4. Redistribution. You may reproduce and distribute copies of the
91
+ Work or Derivative Works thereof in any medium, with or without
92
+ modifications, and in Source or Object form, provided that You
93
+ meet the following conditions:
94
+
95
+ (a) You must give any other recipients of the Work or
96
+ Derivative Works a copy of this License; and
97
+
98
+ (b) You must cause any modified files to carry prominent notices
99
+ stating that You changed the files; and
100
+
101
+ (c) You must retain, in the Source form of any Derivative Works
102
+ that You distribute, all copyright, patent, trademark, and
103
+ attribution notices from the Source form of the Work,
104
+ excluding those notices that do not pertain to any part of
105
+ the Derivative Works; and
106
+
107
+ (d) If the Work includes a "NOTICE" text file as part of its
108
+ distribution, then any Derivative Works that You distribute must
109
+ include a readable copy of the attribution notices contained
110
+ within such NOTICE file, excluding those notices that do not
111
+ pertain to any part of the Derivative Works, in at least one
112
+ of the following places: within a NOTICE text file distributed
113
+ as part of the Derivative Works; within the Source form or
114
+ documentation, if provided along with the Derivative Works; or,
115
+ within a display generated by the Derivative Works, if and
116
+ wherever such third-party notices normally appear. The contents
117
+ of the NOTICE file are for informational purposes only and
118
+ do not modify the License. You may add Your own attribution
119
+ notices within Derivative Works that You distribute, alongside
120
+ or as an addendum to the NOTICE text from the Work, provided
121
+ that such additional attribution notices cannot be construed
122
+ as modifying the License.
123
+
124
+ You may add Your own copyright statement to Your modifications and
125
+ may provide additional or different license terms and conditions
126
+ for use, reproduction, or distribution of Your modifications, or
127
+ for any such Derivative Works as a whole, provided Your use,
128
+ reproduction, and distribution of the Work otherwise complies with
129
+ the conditions stated in this License.
130
+
131
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
132
+ any Contribution intentionally submitted for inclusion in the Work
133
+ by You to the Licensor shall be under the terms and conditions of
134
+ this License, without any additional terms or conditions.
135
+ Notwithstanding the above, nothing herein shall supersede or modify
136
+ the terms of any separate license agreement you may have executed
137
+ with Licensor regarding such Contributions.
138
+
139
+ 6. Trademarks. This License does not grant permission to use the trade
140
+ names, trademarks, service marks, or product names of the Licensor,
141
+ except as required for reasonable and customary use in describing the
142
+ origin of the Work and reproducing the content of the NOTICE file.
143
+
144
+ 7. Disclaimer of Warranty. Unless required by applicable law or
145
+ agreed to in writing, Licensor provides the Work (and each
146
+ Contributor provides its Contributions) on an "AS IS" BASIS,
147
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
148
+ implied, including, without limitation, any warranties or conditions
149
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
150
+ PARTICULAR PURPOSE. You are solely responsible for determining the
151
+ appropriateness of using or redistributing the Work and assume any
152
+ risks associated with Your exercise of permissions under this License.
153
+
154
+ 8. Limitation of Liability. In no event and under no legal theory,
155
+ whether in tort (including negligence), contract, or otherwise,
156
+ unless required by applicable law (such as deliberate and grossly
157
+ negligent acts) or agreed to in writing, shall any Contributor be
158
+ liable to You for damages, including any direct, indirect, special,
159
+ incidental, or consequential damages of any character arising as a
160
+ result of this License or out of the use or inability to use the
161
+ Work (including but not limited to damages for loss of goodwill,
162
+ work stoppage, computer failure or malfunction, or any and all
163
+ other commercial damages or losses), even if such Contributor
164
+ has been advised of the possibility of such damages.
165
+
166
+ 9. Accepting Warranty or Additional Liability. While redistributing
167
+ the Work or Derivative Works thereof, You may choose to offer,
168
+ and charge a fee for, acceptance of support, warranty, indemnity,
169
+ or other liability obligations and/or rights consistent with this
170
+ License. However, in accepting such obligations, You may act only
171
+ on Your own behalf and on Your sole responsibility, not on behalf
172
+ of any other Contributor, and only if You agree to indemnify,
173
+ defend, and hold each Contributor harmless for any liability
174
+ incurred by, or claims asserted against, such Contributor by reason
175
+ of your accepting any such warranty or additional liability.
176
+
177
+ END OF TERMS AND CONDITIONS
178
+
179
+ APPENDIX: How to apply the Apache License to your work.
180
+
181
+ To apply the Apache License to your work, attach the following
182
+ boilerplate notice, with the fields enclosed by brackets "[]"
183
+ replaced with your own identifying information. (Don't include
184
+ the brackets!) The text should be enclosed in the appropriate
185
+ comment syntax for the file format. We also recommend that a
186
+ file or class name and description of purpose be included on the
187
+ same "printed page" as the copyright notice for easier
188
+ identification within third-party archives.
189
+
190
+ Copyright [yyyy] [name of copyright owner]
191
+
192
+ Licensed under the Apache License, Version 2.0 (the "License");
193
+ you may not use this file except in compliance with the License.
194
+ You may obtain a copy of the License at
195
+
196
+ http://www.apache.org/licenses/LICENSE-2.0
197
+
198
+ Unless required by applicable law or agreed to in writing, software
199
+ distributed under the License is distributed on an "AS IS" BASIS,
200
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
201
+ See the License for the specific language governing permissions and
202
+ limitations under the License.
README.flux.upstream.md ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ license: apache-2.0
5
+ tags:
6
+ - text-to-image
7
+ - image-generation
8
+ - flux
9
+ ---
10
+
11
+ ![FLUX.1 [schnell] Grid](./schnell_grid.jpeg)
12
+
13
+ `FLUX.1 [schnell]` is a 12 billion parameter rectified flow transformer capable of generating images from text descriptions.
14
+ For more information, please read our [blog post](https://blackforestlabs.ai/announcing-black-forest-labs/).
15
+
16
+ # Key Features
17
+ 1. Cutting-edge output quality and competitive prompt following, matching the performance of closed source alternatives.
18
+ 2. Trained using latent adversarial diffusion distillation, `FLUX.1 [schnell]` can generate high-quality images in only 1 to 4 steps.
19
+ 3. Released under the `apache-2.0` licence, the model can be used for personal, scientific, and commercial purposes.
20
+
21
+ # Usage
22
+ We provide a reference implementation of `FLUX.1 [schnell]`, as well as sampling code, in a dedicated [github repository](https://github.com/black-forest-labs/flux).
23
+ Developers and creatives looking to build on top of `FLUX.1 [schnell]` are encouraged to use this as a starting point.
24
+
25
+ ## API Endpoints
26
+ The FLUX.1 models are also available via API from the following sources
27
+ - [bfl.ml](https://docs.bfl.ml/) (currently `FLUX.1 [pro]`)
28
+ - [replicate.com](https://replicate.com/collections/flux)
29
+ - [fal.ai](https://fal.ai/models/fal-ai/flux/schnell)
30
+ - [mystic.ai](https://www.mystic.ai/black-forest-labs/flux1-schnell)
31
+
32
+ ## ComfyUI
33
+ `FLUX.1 [schnell]` is also available in [Comfy UI](https://github.com/comfyanonymous/ComfyUI) for local inference with a node-based workflow.
34
+
35
+ ## Diffusers
36
+ To use `FLUX.1 [schnell]` with the 🧨 diffusers python library, first install or upgrade diffusers
37
+
38
+ ```shell
39
+ pip install -U diffusers
40
+ ```
41
+
42
+ Then you can use `FluxPipeline` to run the model
43
+
44
+ ```python
45
+ import torch
46
+ from diffusers import FluxPipeline
47
+
48
+ pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-schnell", torch_dtype=torch.bfloat16)
49
+ pipe.enable_model_cpu_offload() #save some VRAM by offloading the model to CPU. Remove this if you have enough GPU power
50
+
51
+ prompt = "A cat holding a sign that says hello world"
52
+ image = pipe(
53
+ prompt,
54
+ guidance_scale=0.0,
55
+ num_inference_steps=4,
56
+ max_sequence_length=256,
57
+ generator=torch.Generator("cpu").manual_seed(0)
58
+ ).images[0]
59
+ image.save("flux-schnell.png")
60
+ ```
61
+
62
+ To learn more check out the [diffusers](https://huggingface.co/docs/diffusers/main/en/api/pipelines/flux) documentation
63
+
64
+ ---
65
+ # Limitations
66
+ - This model is not intended or able to provide factual information.
67
+ - As a statistical model this checkpoint might amplify existing societal biases.
68
+ - The model may fail to generate output that matches the prompts.
69
+ - Prompt following is heavily influenced by the prompting-style.
70
+
71
+ # Out-of-Scope Use
72
+ The model and its derivatives may not be used
73
+
74
+ - In any way that violates any applicable national, federal, state, local or international law or regulation.
75
+ - For the purpose of exploiting, harming or attempting to exploit or harm minors in any way; including but not limited to the solicitation, creation, acquisition, or dissemination of child exploitative content.
76
+ - To generate or disseminate verifiably false information and/or content with the purpose of harming others.
77
+ - To generate or disseminate personal identifiable information that can be used to harm an individual.
78
+ - To harass, abuse, threaten, stalk, or bully individuals or groups of individuals.
79
+ - To create non-consensual nudity or illegal pornographic content.
80
+ - For fully automated decision making that adversely impacts an individual's legal rights or otherwise creates or modifies a binding, enforceable obligation.
81
+ - Generating or facilitating large-scale disinformation campaigns.
README.md ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # altbit/avatarsvc-prep-models
2
+
3
+ Unmodified re-publication of two Apache-2.0 models, as sub-folders of one repository, so that one model reference caches both (the avatar service's prep endpoint). **No weight, config or tokenizer file has been changed.** Each sub-folder is an unmodified copy (or, for a diffusers pipeline, an unmodified SUBSET: the components its `model_index.json` names) of the upstream commit below.
4
+
5
+ | Folder | Upstream (repo @ commit) | Files | Licence |
6
+ |---|---|---|---|
7
+ | `flux/` | [black-forest-labs/FLUX.1-schnell](https://huggingface.co/black-forest-labs/FLUX.1-schnell) @ `741f7c3ce8b383c54771c7003378a50191e9efe9` | 23 | Apache-2.0 |
8
+ | `qwen3-tts/` | [Qwen/Qwen3-TTS-12Hz-1.7B-Base](https://huggingface.co/Qwen/Qwen3-TTS-12Hz-1.7B-Base) @ `fd4b254389122332181a7c3db7f27e918eec64e3` | 13 | Apache-2.0 |
9
+
10
+ Upstream model cards are copied beside this file where the sub-folder does not already carry one (`README.<folder>.upstream.md`). Full licence: `LICENSE`.
11
+
12
+ ## Files (sha256)
13
+
14
+ - `flux/model_index.json` 536 B `24946df21ff25e210486b5f6b14208983a90c9c73f8d48cfa724c0e4e03f7201`
15
+ - `flux/scheduler/scheduler_config.json` 274 B `b129cebacf8f851867ec5c7c4d3f4bf787e232525a53becf4df5a72278a788d5`
16
+ - `flux/text_encoder/config.json` 613 B `d79d5c8c6ce85112a923d621a5412886ddbbb0636210fc0f72f450582e675542`
17
+ - `flux/text_encoder/model.safetensors` 246144352 B `893d67a23f4693ed42cdab4cbad7fe3e727cf59609c40da28a46b5470f9ed082`
18
+ - `flux/text_encoder_2/config.json` 782 B `9001e5a8ae0571a362f806b87b6105dd1a15c33dca237b606d2561164109beeb`
19
+ - `flux/text_encoder_2/model-00001-of-00002.safetensors` 4994582224 B `ec87bffd1923e8b2774a6d240c922a41f6143081d52cf83b8fe39e9d838c893e`
20
+ - `flux/text_encoder_2/model-00002-of-00002.safetensors` 4530066360 B `a5640855b301fcdbceddfa90ae8066cd9414aff020552a201a255ecf2059da00`
21
+ - `flux/text_encoder_2/model.safetensors.index.json` 19885 B `3bacec0f0cf392399d4a385908f67dd73df99c9e9cfee669f148858ba9fbdb0a`
22
+ - `flux/tokenizer/merges.txt` 524619 B `9fd691f7c8039210e0fced15865466c65820d09b63988b0174bfe25de299051a`
23
+ - `flux/tokenizer/special_tokens_map.json` 588 B `2cdb3b8331a60c92fc1e55a13e9fd61fd2293c5a51275fdcccd62b780052530e`
24
+ - `flux/tokenizer/tokenizer_config.json` 705 B `6bdcee9ccce2a16ca2b4c0c5ed00b42c50ea225f4472a8c4c1e963a2902c2881`
25
+ - `flux/tokenizer/vocab.json` 1059962 B `e089ad92ba36837a0d31433e555c8f45fe601ab5c221d4f607ded32d9f7a4349`
26
+ - `flux/tokenizer_2/special_tokens_map.json` 2543 B `7a1985a994c41886db38c719d2a3d2f40606663cc19d7c5d6a85d349320e06d2`
27
+ - `flux/tokenizer_2/spiece.model` 791656 B `d60acb128cf7b7f2536e8f38a5b18a05535c9e14c7a355904270e15b0945ea86`
28
+ - `flux/tokenizer_2/tokenizer.json` 2424235 B `f5dfec163765e18e270537fe896c49f5fad74db1525641d9b255a3008b999596`
29
+ - `flux/tokenizer_2/tokenizer_config.json` 20817 B `1a3d2db64215ed77854dd4208aac5f8361c1b5471cabd19c0ef1472d1a895eb0`
30
+ - `flux/transformer/config.json` 321 B `397cfb92299488013ec3af6142a2a877366f8d2e44efbb4f3e33479e7960d3d0`
31
+ - `flux/transformer/diffusion_pytorch_model-00001-of-00003.safetensors` 9962580296 B `9b633dbe87316385c5b1c262bd4b5a01e3d955170661d63dcec8a01e89c0d820`
32
+ - `flux/transformer/diffusion_pytorch_model-00002-of-00003.safetensors` 9949328904 B `58b4434078f0c2567ddc54e3b5cbf39626ab55fbd9d5c22956e183668f535dec`
33
+ - `flux/transformer/diffusion_pytorch_model-00003-of-00003.safetensors` 3870584832 B `e2cbc25471ed5186e69a9b51098300cb2f612556453e38a372c851a220ed238d`
34
+ - `flux/transformer/diffusion_pytorch_model.safetensors.index.json` 120822 B `783f857a5872f069e75daf4a5abe5efd6ff9ec2f37d71159767910cebfe048a6`
35
+ - `flux/vae/config.json` 774 B `bc1e208f414a315365fbecf426838f43b87c9d5c051219e0968a56e2644b2998`
36
+ - `flux/vae/diffusion_pytorch_model.safetensors` 167666902 B `f5b59a26851551b67ae1fe58d32e76486e1e812def4696a4bea97f16604d40a3`
37
+ - `qwen3-tts/.gitattributes` 1519 B `11ad7efa24975ee4b0c3c3a38ed18737f0658a5f75a0a96787b576a78a023361`
38
+ - `qwen3-tts/README.md` 57817 B `621b7e88f1867bc03d817176fc1f7f55f4b5a70654b2a07fdcda1e169efb024b`
39
+ - `qwen3-tts/config.json` 4494 B `b4f01752d15a488abde3e1ab44723ae4f4b9e68a4037257b098b3737893cc1f9`
40
+ - `qwen3-tts/generation_config.json` 245 B `f1b90b4513f3b34c62851049e2492d7b4c5940daf1276f89c82b8ef04127f3aa`
41
+ - `qwen3-tts/merges.txt` 1671839 B `599bab54075088774b1733fde865d5bd747cbcc7a547c5bc12610e874e26f5e3`
42
+ - `qwen3-tts/model.safetensors` 3857413744 B `38fc7fc51c5e776e840414b6fd443962e9411b9654888fd7913e4da643cb857c`
43
+ - `qwen3-tts/preprocessor_config.json` 127 B `efdde1022ea9d76928bf7a9cd53139138f5ba2e466e837f08f6105ab1af1c119`
44
+ - `qwen3-tts/speech_tokenizer/config.json` 2336 B `ee65bb901c876664ab8707c487157aa1a6ee57c65969b28fb5ec9dc211e68167`
45
+ - `qwen3-tts/speech_tokenizer/configuration.json` 76 B `6bc26d64eb5024b4d1dab5a52371958b429256d6c9d59787f1f5294a54e0cebd`
46
+ - `qwen3-tts/speech_tokenizer/model.safetensors` 682293092 B `836b7b357f5ea43e889936a3709af68dfe3751881acefe4ecf0dbd30ba571258`
47
+ - `qwen3-tts/speech_tokenizer/preprocessor_config.json` 234 B `fcb3805e597e786d4067706e602f6688524640f8d3396790e2e09b5942fcbdfb`
48
+ - `qwen3-tts/tokenizer_config.json` 7344 B `dc3c31c3bdaedd5016382bb3cbe07323026775ad51f5a4fb564505992ae4a670`
49
+ - `qwen3-tts/vocab.json` 2776833 B `ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910`
flux/model_index.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_class_name": "FluxPipeline",
3
+ "_diffusers_version": "0.30.0.dev0",
4
+ "scheduler": [
5
+ "diffusers",
6
+ "FlowMatchEulerDiscreteScheduler"
7
+ ],
8
+ "text_encoder": [
9
+ "transformers",
10
+ "CLIPTextModel"
11
+ ],
12
+ "text_encoder_2": [
13
+ "transformers",
14
+ "T5EncoderModel"
15
+ ],
16
+ "tokenizer": [
17
+ "transformers",
18
+ "CLIPTokenizer"
19
+ ],
20
+ "tokenizer_2": [
21
+ "transformers",
22
+ "T5TokenizerFast"
23
+ ],
24
+ "transformer": [
25
+ "diffusers",
26
+ "FluxTransformer2DModel"
27
+ ],
28
+ "vae": [
29
+ "diffusers",
30
+ "AutoencoderKL"
31
+ ]
32
+ }
flux/scheduler/scheduler_config.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_class_name": "FlowMatchEulerDiscreteScheduler",
3
+ "_diffusers_version": "0.30.0.dev0",
4
+ "base_image_seq_len": 256,
5
+ "base_shift": 0.5,
6
+ "max_image_seq_len": 4096,
7
+ "max_shift": 1.15,
8
+ "num_train_timesteps": 1000,
9
+ "shift": 1.0,
10
+ "use_dynamic_shifting": false
11
+ }
flux/text_encoder/config.json ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/clip-vit-large-patch14",
3
+ "architectures": [
4
+ "CLIPTextModel"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 0,
8
+ "dropout": 0.0,
9
+ "eos_token_id": 2,
10
+ "hidden_act": "quick_gelu",
11
+ "hidden_size": 768,
12
+ "initializer_factor": 1.0,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 3072,
15
+ "layer_norm_eps": 1e-05,
16
+ "max_position_embeddings": 77,
17
+ "model_type": "clip_text_model",
18
+ "num_attention_heads": 12,
19
+ "num_hidden_layers": 12,
20
+ "pad_token_id": 1,
21
+ "projection_dim": 768,
22
+ "torch_dtype": "bfloat16",
23
+ "transformers_version": "4.43.3",
24
+ "vocab_size": 49408
25
+ }
flux/text_encoder/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:893d67a23f4693ed42cdab4cbad7fe3e727cf59609c40da28a46b5470f9ed082
3
+ size 246144352
flux/text_encoder_2/config.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "google/t5-v1_1-xxl",
3
+ "architectures": [
4
+ "T5EncoderModel"
5
+ ],
6
+ "classifier_dropout": 0.0,
7
+ "d_ff": 10240,
8
+ "d_kv": 64,
9
+ "d_model": 4096,
10
+ "decoder_start_token_id": 0,
11
+ "dense_act_fn": "gelu_new",
12
+ "dropout_rate": 0.1,
13
+ "eos_token_id": 1,
14
+ "feed_forward_proj": "gated-gelu",
15
+ "initializer_factor": 1.0,
16
+ "is_encoder_decoder": true,
17
+ "is_gated_act": true,
18
+ "layer_norm_epsilon": 1e-06,
19
+ "model_type": "t5",
20
+ "num_decoder_layers": 24,
21
+ "num_heads": 64,
22
+ "num_layers": 24,
23
+ "output_past": true,
24
+ "pad_token_id": 0,
25
+ "relative_attention_max_distance": 128,
26
+ "relative_attention_num_buckets": 32,
27
+ "tie_word_embeddings": false,
28
+ "torch_dtype": "bfloat16",
29
+ "transformers_version": "4.43.3",
30
+ "use_cache": true,
31
+ "vocab_size": 32128
32
+ }
flux/text_encoder_2/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ec87bffd1923e8b2774a6d240c922a41f6143081d52cf83b8fe39e9d838c893e
3
+ size 4994582224
flux/text_encoder_2/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a5640855b301fcdbceddfa90ae8066cd9414aff020552a201a255ecf2059da00
3
+ size 4530066360
flux/text_encoder_2/model.safetensors.index.json ADDED
@@ -0,0 +1,226 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 9524621312
4
+ },
5
+ "weight_map": {
6
+ "encoder.block.0.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
7
+ "encoder.block.0.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
8
+ "encoder.block.0.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
9
+ "encoder.block.0.layer.0.SelfAttention.relative_attention_bias.weight": "model-00001-of-00002.safetensors",
10
+ "encoder.block.0.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
11
+ "encoder.block.0.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
12
+ "encoder.block.0.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
13
+ "encoder.block.0.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
14
+ "encoder.block.0.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
15
+ "encoder.block.0.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
16
+ "encoder.block.1.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
17
+ "encoder.block.1.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
18
+ "encoder.block.1.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
19
+ "encoder.block.1.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
20
+ "encoder.block.1.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
21
+ "encoder.block.1.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
22
+ "encoder.block.1.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
23
+ "encoder.block.1.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
24
+ "encoder.block.1.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
25
+ "encoder.block.10.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
26
+ "encoder.block.10.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
27
+ "encoder.block.10.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
28
+ "encoder.block.10.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
29
+ "encoder.block.10.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
30
+ "encoder.block.10.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
31
+ "encoder.block.10.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
32
+ "encoder.block.10.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
33
+ "encoder.block.10.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
34
+ "encoder.block.11.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
35
+ "encoder.block.11.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
36
+ "encoder.block.11.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
37
+ "encoder.block.11.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
38
+ "encoder.block.11.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
39
+ "encoder.block.11.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
40
+ "encoder.block.11.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
41
+ "encoder.block.11.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
42
+ "encoder.block.11.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
43
+ "encoder.block.12.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
44
+ "encoder.block.12.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
45
+ "encoder.block.12.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
46
+ "encoder.block.12.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
47
+ "encoder.block.12.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
48
+ "encoder.block.12.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
49
+ "encoder.block.12.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
50
+ "encoder.block.12.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
51
+ "encoder.block.12.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
52
+ "encoder.block.13.layer.0.SelfAttention.k.weight": "model-00002-of-00002.safetensors",
53
+ "encoder.block.13.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
54
+ "encoder.block.13.layer.0.SelfAttention.q.weight": "model-00002-of-00002.safetensors",
55
+ "encoder.block.13.layer.0.SelfAttention.v.weight": "model-00002-of-00002.safetensors",
56
+ "encoder.block.13.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
57
+ "encoder.block.13.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
58
+ "encoder.block.13.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
59
+ "encoder.block.13.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
60
+ "encoder.block.13.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
61
+ "encoder.block.14.layer.0.SelfAttention.k.weight": "model-00002-of-00002.safetensors",
62
+ "encoder.block.14.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
63
+ "encoder.block.14.layer.0.SelfAttention.q.weight": "model-00002-of-00002.safetensors",
64
+ "encoder.block.14.layer.0.SelfAttention.v.weight": "model-00002-of-00002.safetensors",
65
+ "encoder.block.14.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
66
+ "encoder.block.14.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
67
+ "encoder.block.14.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
68
+ "encoder.block.14.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
69
+ "encoder.block.14.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
70
+ "encoder.block.15.layer.0.SelfAttention.k.weight": "model-00002-of-00002.safetensors",
71
+ "encoder.block.15.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
72
+ "encoder.block.15.layer.0.SelfAttention.q.weight": "model-00002-of-00002.safetensors",
73
+ "encoder.block.15.layer.0.SelfAttention.v.weight": "model-00002-of-00002.safetensors",
74
+ "encoder.block.15.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
75
+ "encoder.block.15.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
76
+ "encoder.block.15.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
77
+ "encoder.block.15.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
78
+ "encoder.block.15.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
79
+ "encoder.block.16.layer.0.SelfAttention.k.weight": "model-00002-of-00002.safetensors",
80
+ "encoder.block.16.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
81
+ "encoder.block.16.layer.0.SelfAttention.q.weight": "model-00002-of-00002.safetensors",
82
+ "encoder.block.16.layer.0.SelfAttention.v.weight": "model-00002-of-00002.safetensors",
83
+ "encoder.block.16.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
84
+ "encoder.block.16.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
85
+ "encoder.block.16.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
86
+ "encoder.block.16.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
87
+ "encoder.block.16.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
88
+ "encoder.block.17.layer.0.SelfAttention.k.weight": "model-00002-of-00002.safetensors",
89
+ "encoder.block.17.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
90
+ "encoder.block.17.layer.0.SelfAttention.q.weight": "model-00002-of-00002.safetensors",
91
+ "encoder.block.17.layer.0.SelfAttention.v.weight": "model-00002-of-00002.safetensors",
92
+ "encoder.block.17.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
93
+ "encoder.block.17.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
94
+ "encoder.block.17.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
95
+ "encoder.block.17.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
96
+ "encoder.block.17.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
97
+ "encoder.block.18.layer.0.SelfAttention.k.weight": "model-00002-of-00002.safetensors",
98
+ "encoder.block.18.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
99
+ "encoder.block.18.layer.0.SelfAttention.q.weight": "model-00002-of-00002.safetensors",
100
+ "encoder.block.18.layer.0.SelfAttention.v.weight": "model-00002-of-00002.safetensors",
101
+ "encoder.block.18.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
102
+ "encoder.block.18.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
103
+ "encoder.block.18.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
104
+ "encoder.block.18.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
105
+ "encoder.block.18.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
106
+ "encoder.block.19.layer.0.SelfAttention.k.weight": "model-00002-of-00002.safetensors",
107
+ "encoder.block.19.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
108
+ "encoder.block.19.layer.0.SelfAttention.q.weight": "model-00002-of-00002.safetensors",
109
+ "encoder.block.19.layer.0.SelfAttention.v.weight": "model-00002-of-00002.safetensors",
110
+ "encoder.block.19.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
111
+ "encoder.block.19.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
112
+ "encoder.block.19.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
113
+ "encoder.block.19.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
114
+ "encoder.block.19.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
115
+ "encoder.block.2.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
116
+ "encoder.block.2.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
117
+ "encoder.block.2.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
118
+ "encoder.block.2.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
119
+ "encoder.block.2.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
120
+ "encoder.block.2.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
121
+ "encoder.block.2.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
122
+ "encoder.block.2.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
123
+ "encoder.block.2.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
124
+ "encoder.block.20.layer.0.SelfAttention.k.weight": "model-00002-of-00002.safetensors",
125
+ "encoder.block.20.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
126
+ "encoder.block.20.layer.0.SelfAttention.q.weight": "model-00002-of-00002.safetensors",
127
+ "encoder.block.20.layer.0.SelfAttention.v.weight": "model-00002-of-00002.safetensors",
128
+ "encoder.block.20.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
129
+ "encoder.block.20.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
130
+ "encoder.block.20.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
131
+ "encoder.block.20.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
132
+ "encoder.block.20.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
133
+ "encoder.block.21.layer.0.SelfAttention.k.weight": "model-00002-of-00002.safetensors",
134
+ "encoder.block.21.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
135
+ "encoder.block.21.layer.0.SelfAttention.q.weight": "model-00002-of-00002.safetensors",
136
+ "encoder.block.21.layer.0.SelfAttention.v.weight": "model-00002-of-00002.safetensors",
137
+ "encoder.block.21.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
138
+ "encoder.block.21.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
139
+ "encoder.block.21.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
140
+ "encoder.block.21.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
141
+ "encoder.block.21.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
142
+ "encoder.block.22.layer.0.SelfAttention.k.weight": "model-00002-of-00002.safetensors",
143
+ "encoder.block.22.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
144
+ "encoder.block.22.layer.0.SelfAttention.q.weight": "model-00002-of-00002.safetensors",
145
+ "encoder.block.22.layer.0.SelfAttention.v.weight": "model-00002-of-00002.safetensors",
146
+ "encoder.block.22.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
147
+ "encoder.block.22.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
148
+ "encoder.block.22.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
149
+ "encoder.block.22.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
150
+ "encoder.block.22.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
151
+ "encoder.block.23.layer.0.SelfAttention.k.weight": "model-00002-of-00002.safetensors",
152
+ "encoder.block.23.layer.0.SelfAttention.o.weight": "model-00002-of-00002.safetensors",
153
+ "encoder.block.23.layer.0.SelfAttention.q.weight": "model-00002-of-00002.safetensors",
154
+ "encoder.block.23.layer.0.SelfAttention.v.weight": "model-00002-of-00002.safetensors",
155
+ "encoder.block.23.layer.0.layer_norm.weight": "model-00002-of-00002.safetensors",
156
+ "encoder.block.23.layer.1.DenseReluDense.wi_0.weight": "model-00002-of-00002.safetensors",
157
+ "encoder.block.23.layer.1.DenseReluDense.wi_1.weight": "model-00002-of-00002.safetensors",
158
+ "encoder.block.23.layer.1.DenseReluDense.wo.weight": "model-00002-of-00002.safetensors",
159
+ "encoder.block.23.layer.1.layer_norm.weight": "model-00002-of-00002.safetensors",
160
+ "encoder.block.3.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
161
+ "encoder.block.3.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
162
+ "encoder.block.3.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
163
+ "encoder.block.3.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
164
+ "encoder.block.3.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
165
+ "encoder.block.3.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
166
+ "encoder.block.3.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
167
+ "encoder.block.3.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
168
+ "encoder.block.3.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
169
+ "encoder.block.4.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
170
+ "encoder.block.4.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
171
+ "encoder.block.4.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
172
+ "encoder.block.4.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
173
+ "encoder.block.4.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
174
+ "encoder.block.4.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
175
+ "encoder.block.4.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
176
+ "encoder.block.4.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
177
+ "encoder.block.4.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
178
+ "encoder.block.5.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
179
+ "encoder.block.5.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
180
+ "encoder.block.5.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
181
+ "encoder.block.5.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
182
+ "encoder.block.5.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
183
+ "encoder.block.5.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
184
+ "encoder.block.5.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
185
+ "encoder.block.5.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
186
+ "encoder.block.5.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
187
+ "encoder.block.6.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
188
+ "encoder.block.6.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
189
+ "encoder.block.6.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
190
+ "encoder.block.6.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
191
+ "encoder.block.6.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
192
+ "encoder.block.6.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
193
+ "encoder.block.6.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
194
+ "encoder.block.6.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
195
+ "encoder.block.6.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
196
+ "encoder.block.7.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
197
+ "encoder.block.7.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
198
+ "encoder.block.7.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
199
+ "encoder.block.7.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
200
+ "encoder.block.7.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
201
+ "encoder.block.7.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
202
+ "encoder.block.7.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
203
+ "encoder.block.7.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
204
+ "encoder.block.7.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
205
+ "encoder.block.8.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
206
+ "encoder.block.8.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
207
+ "encoder.block.8.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
208
+ "encoder.block.8.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
209
+ "encoder.block.8.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
210
+ "encoder.block.8.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
211
+ "encoder.block.8.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
212
+ "encoder.block.8.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
213
+ "encoder.block.8.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
214
+ "encoder.block.9.layer.0.SelfAttention.k.weight": "model-00001-of-00002.safetensors",
215
+ "encoder.block.9.layer.0.SelfAttention.o.weight": "model-00001-of-00002.safetensors",
216
+ "encoder.block.9.layer.0.SelfAttention.q.weight": "model-00001-of-00002.safetensors",
217
+ "encoder.block.9.layer.0.SelfAttention.v.weight": "model-00001-of-00002.safetensors",
218
+ "encoder.block.9.layer.0.layer_norm.weight": "model-00001-of-00002.safetensors",
219
+ "encoder.block.9.layer.1.DenseReluDense.wi_0.weight": "model-00001-of-00002.safetensors",
220
+ "encoder.block.9.layer.1.DenseReluDense.wi_1.weight": "model-00001-of-00002.safetensors",
221
+ "encoder.block.9.layer.1.DenseReluDense.wo.weight": "model-00001-of-00002.safetensors",
222
+ "encoder.block.9.layer.1.layer_norm.weight": "model-00001-of-00002.safetensors",
223
+ "encoder.final_layer_norm.weight": "model-00002-of-00002.safetensors",
224
+ "shared.weight": "model-00001-of-00002.safetensors"
225
+ }
226
+ }
flux/tokenizer/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
flux/tokenizer/special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<|startoftext|>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<|endoftext|>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "<|endoftext|>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "<|endoftext|>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
flux/tokenizer/tokenizer_config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "49406": {
5
+ "content": "<|startoftext|>",
6
+ "lstrip": false,
7
+ "normalized": true,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ },
12
+ "49407": {
13
+ "content": "<|endoftext|>",
14
+ "lstrip": false,
15
+ "normalized": false,
16
+ "rstrip": false,
17
+ "single_word": false,
18
+ "special": true
19
+ }
20
+ },
21
+ "bos_token": "<|startoftext|>",
22
+ "clean_up_tokenization_spaces": true,
23
+ "do_lower_case": true,
24
+ "eos_token": "<|endoftext|>",
25
+ "errors": "replace",
26
+ "model_max_length": 77,
27
+ "pad_token": "<|endoftext|>",
28
+ "tokenizer_class": "CLIPTokenizer",
29
+ "unk_token": "<|endoftext|>"
30
+ }
flux/tokenizer/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
flux/tokenizer_2/special_tokens_map.json ADDED
@@ -0,0 +1,125 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<extra_id_0>",
4
+ "<extra_id_1>",
5
+ "<extra_id_2>",
6
+ "<extra_id_3>",
7
+ "<extra_id_4>",
8
+ "<extra_id_5>",
9
+ "<extra_id_6>",
10
+ "<extra_id_7>",
11
+ "<extra_id_8>",
12
+ "<extra_id_9>",
13
+ "<extra_id_10>",
14
+ "<extra_id_11>",
15
+ "<extra_id_12>",
16
+ "<extra_id_13>",
17
+ "<extra_id_14>",
18
+ "<extra_id_15>",
19
+ "<extra_id_16>",
20
+ "<extra_id_17>",
21
+ "<extra_id_18>",
22
+ "<extra_id_19>",
23
+ "<extra_id_20>",
24
+ "<extra_id_21>",
25
+ "<extra_id_22>",
26
+ "<extra_id_23>",
27
+ "<extra_id_24>",
28
+ "<extra_id_25>",
29
+ "<extra_id_26>",
30
+ "<extra_id_27>",
31
+ "<extra_id_28>",
32
+ "<extra_id_29>",
33
+ "<extra_id_30>",
34
+ "<extra_id_31>",
35
+ "<extra_id_32>",
36
+ "<extra_id_33>",
37
+ "<extra_id_34>",
38
+ "<extra_id_35>",
39
+ "<extra_id_36>",
40
+ "<extra_id_37>",
41
+ "<extra_id_38>",
42
+ "<extra_id_39>",
43
+ "<extra_id_40>",
44
+ "<extra_id_41>",
45
+ "<extra_id_42>",
46
+ "<extra_id_43>",
47
+ "<extra_id_44>",
48
+ "<extra_id_45>",
49
+ "<extra_id_46>",
50
+ "<extra_id_47>",
51
+ "<extra_id_48>",
52
+ "<extra_id_49>",
53
+ "<extra_id_50>",
54
+ "<extra_id_51>",
55
+ "<extra_id_52>",
56
+ "<extra_id_53>",
57
+ "<extra_id_54>",
58
+ "<extra_id_55>",
59
+ "<extra_id_56>",
60
+ "<extra_id_57>",
61
+ "<extra_id_58>",
62
+ "<extra_id_59>",
63
+ "<extra_id_60>",
64
+ "<extra_id_61>",
65
+ "<extra_id_62>",
66
+ "<extra_id_63>",
67
+ "<extra_id_64>",
68
+ "<extra_id_65>",
69
+ "<extra_id_66>",
70
+ "<extra_id_67>",
71
+ "<extra_id_68>",
72
+ "<extra_id_69>",
73
+ "<extra_id_70>",
74
+ "<extra_id_71>",
75
+ "<extra_id_72>",
76
+ "<extra_id_73>",
77
+ "<extra_id_74>",
78
+ "<extra_id_75>",
79
+ "<extra_id_76>",
80
+ "<extra_id_77>",
81
+ "<extra_id_78>",
82
+ "<extra_id_79>",
83
+ "<extra_id_80>",
84
+ "<extra_id_81>",
85
+ "<extra_id_82>",
86
+ "<extra_id_83>",
87
+ "<extra_id_84>",
88
+ "<extra_id_85>",
89
+ "<extra_id_86>",
90
+ "<extra_id_87>",
91
+ "<extra_id_88>",
92
+ "<extra_id_89>",
93
+ "<extra_id_90>",
94
+ "<extra_id_91>",
95
+ "<extra_id_92>",
96
+ "<extra_id_93>",
97
+ "<extra_id_94>",
98
+ "<extra_id_95>",
99
+ "<extra_id_96>",
100
+ "<extra_id_97>",
101
+ "<extra_id_98>",
102
+ "<extra_id_99>"
103
+ ],
104
+ "eos_token": {
105
+ "content": "</s>",
106
+ "lstrip": false,
107
+ "normalized": false,
108
+ "rstrip": false,
109
+ "single_word": false
110
+ },
111
+ "pad_token": {
112
+ "content": "<pad>",
113
+ "lstrip": false,
114
+ "normalized": false,
115
+ "rstrip": false,
116
+ "single_word": false
117
+ },
118
+ "unk_token": {
119
+ "content": "<unk>",
120
+ "lstrip": false,
121
+ "normalized": false,
122
+ "rstrip": false,
123
+ "single_word": false
124
+ }
125
+ }
flux/tokenizer_2/spiece.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d60acb128cf7b7f2536e8f38a5b18a05535c9e14c7a355904270e15b0945ea86
3
+ size 791656
flux/tokenizer_2/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
flux/tokenizer_2/tokenizer_config.json ADDED
@@ -0,0 +1,940 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": true,
3
+ "added_tokens_decoder": {
4
+ "0": {
5
+ "content": "<pad>",
6
+ "lstrip": false,
7
+ "normalized": false,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ },
12
+ "1": {
13
+ "content": "</s>",
14
+ "lstrip": false,
15
+ "normalized": false,
16
+ "rstrip": false,
17
+ "single_word": false,
18
+ "special": true
19
+ },
20
+ "2": {
21
+ "content": "<unk>",
22
+ "lstrip": false,
23
+ "normalized": false,
24
+ "rstrip": false,
25
+ "single_word": false,
26
+ "special": true
27
+ },
28
+ "32000": {
29
+ "content": "<extra_id_99>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false,
34
+ "special": true
35
+ },
36
+ "32001": {
37
+ "content": "<extra_id_98>",
38
+ "lstrip": false,
39
+ "normalized": false,
40
+ "rstrip": false,
41
+ "single_word": false,
42
+ "special": true
43
+ },
44
+ "32002": {
45
+ "content": "<extra_id_97>",
46
+ "lstrip": false,
47
+ "normalized": false,
48
+ "rstrip": false,
49
+ "single_word": false,
50
+ "special": true
51
+ },
52
+ "32003": {
53
+ "content": "<extra_id_96>",
54
+ "lstrip": false,
55
+ "normalized": false,
56
+ "rstrip": false,
57
+ "single_word": false,
58
+ "special": true
59
+ },
60
+ "32004": {
61
+ "content": "<extra_id_95>",
62
+ "lstrip": false,
63
+ "normalized": false,
64
+ "rstrip": false,
65
+ "single_word": false,
66
+ "special": true
67
+ },
68
+ "32005": {
69
+ "content": "<extra_id_94>",
70
+ "lstrip": false,
71
+ "normalized": false,
72
+ "rstrip": false,
73
+ "single_word": false,
74
+ "special": true
75
+ },
76
+ "32006": {
77
+ "content": "<extra_id_93>",
78
+ "lstrip": false,
79
+ "normalized": false,
80
+ "rstrip": false,
81
+ "single_word": false,
82
+ "special": true
83
+ },
84
+ "32007": {
85
+ "content": "<extra_id_92>",
86
+ "lstrip": false,
87
+ "normalized": false,
88
+ "rstrip": false,
89
+ "single_word": false,
90
+ "special": true
91
+ },
92
+ "32008": {
93
+ "content": "<extra_id_91>",
94
+ "lstrip": false,
95
+ "normalized": false,
96
+ "rstrip": false,
97
+ "single_word": false,
98
+ "special": true
99
+ },
100
+ "32009": {
101
+ "content": "<extra_id_90>",
102
+ "lstrip": false,
103
+ "normalized": false,
104
+ "rstrip": false,
105
+ "single_word": false,
106
+ "special": true
107
+ },
108
+ "32010": {
109
+ "content": "<extra_id_89>",
110
+ "lstrip": false,
111
+ "normalized": false,
112
+ "rstrip": false,
113
+ "single_word": false,
114
+ "special": true
115
+ },
116
+ "32011": {
117
+ "content": "<extra_id_88>",
118
+ "lstrip": false,
119
+ "normalized": false,
120
+ "rstrip": false,
121
+ "single_word": false,
122
+ "special": true
123
+ },
124
+ "32012": {
125
+ "content": "<extra_id_87>",
126
+ "lstrip": false,
127
+ "normalized": false,
128
+ "rstrip": false,
129
+ "single_word": false,
130
+ "special": true
131
+ },
132
+ "32013": {
133
+ "content": "<extra_id_86>",
134
+ "lstrip": false,
135
+ "normalized": false,
136
+ "rstrip": false,
137
+ "single_word": false,
138
+ "special": true
139
+ },
140
+ "32014": {
141
+ "content": "<extra_id_85>",
142
+ "lstrip": false,
143
+ "normalized": false,
144
+ "rstrip": false,
145
+ "single_word": false,
146
+ "special": true
147
+ },
148
+ "32015": {
149
+ "content": "<extra_id_84>",
150
+ "lstrip": false,
151
+ "normalized": false,
152
+ "rstrip": false,
153
+ "single_word": false,
154
+ "special": true
155
+ },
156
+ "32016": {
157
+ "content": "<extra_id_83>",
158
+ "lstrip": false,
159
+ "normalized": false,
160
+ "rstrip": false,
161
+ "single_word": false,
162
+ "special": true
163
+ },
164
+ "32017": {
165
+ "content": "<extra_id_82>",
166
+ "lstrip": false,
167
+ "normalized": false,
168
+ "rstrip": false,
169
+ "single_word": false,
170
+ "special": true
171
+ },
172
+ "32018": {
173
+ "content": "<extra_id_81>",
174
+ "lstrip": false,
175
+ "normalized": false,
176
+ "rstrip": false,
177
+ "single_word": false,
178
+ "special": true
179
+ },
180
+ "32019": {
181
+ "content": "<extra_id_80>",
182
+ "lstrip": false,
183
+ "normalized": false,
184
+ "rstrip": false,
185
+ "single_word": false,
186
+ "special": true
187
+ },
188
+ "32020": {
189
+ "content": "<extra_id_79>",
190
+ "lstrip": false,
191
+ "normalized": false,
192
+ "rstrip": false,
193
+ "single_word": false,
194
+ "special": true
195
+ },
196
+ "32021": {
197
+ "content": "<extra_id_78>",
198
+ "lstrip": false,
199
+ "normalized": false,
200
+ "rstrip": false,
201
+ "single_word": false,
202
+ "special": true
203
+ },
204
+ "32022": {
205
+ "content": "<extra_id_77>",
206
+ "lstrip": false,
207
+ "normalized": false,
208
+ "rstrip": false,
209
+ "single_word": false,
210
+ "special": true
211
+ },
212
+ "32023": {
213
+ "content": "<extra_id_76>",
214
+ "lstrip": false,
215
+ "normalized": false,
216
+ "rstrip": false,
217
+ "single_word": false,
218
+ "special": true
219
+ },
220
+ "32024": {
221
+ "content": "<extra_id_75>",
222
+ "lstrip": false,
223
+ "normalized": false,
224
+ "rstrip": false,
225
+ "single_word": false,
226
+ "special": true
227
+ },
228
+ "32025": {
229
+ "content": "<extra_id_74>",
230
+ "lstrip": false,
231
+ "normalized": false,
232
+ "rstrip": false,
233
+ "single_word": false,
234
+ "special": true
235
+ },
236
+ "32026": {
237
+ "content": "<extra_id_73>",
238
+ "lstrip": false,
239
+ "normalized": false,
240
+ "rstrip": false,
241
+ "single_word": false,
242
+ "special": true
243
+ },
244
+ "32027": {
245
+ "content": "<extra_id_72>",
246
+ "lstrip": false,
247
+ "normalized": false,
248
+ "rstrip": false,
249
+ "single_word": false,
250
+ "special": true
251
+ },
252
+ "32028": {
253
+ "content": "<extra_id_71>",
254
+ "lstrip": false,
255
+ "normalized": false,
256
+ "rstrip": false,
257
+ "single_word": false,
258
+ "special": true
259
+ },
260
+ "32029": {
261
+ "content": "<extra_id_70>",
262
+ "lstrip": false,
263
+ "normalized": false,
264
+ "rstrip": false,
265
+ "single_word": false,
266
+ "special": true
267
+ },
268
+ "32030": {
269
+ "content": "<extra_id_69>",
270
+ "lstrip": false,
271
+ "normalized": false,
272
+ "rstrip": false,
273
+ "single_word": false,
274
+ "special": true
275
+ },
276
+ "32031": {
277
+ "content": "<extra_id_68>",
278
+ "lstrip": false,
279
+ "normalized": false,
280
+ "rstrip": false,
281
+ "single_word": false,
282
+ "special": true
283
+ },
284
+ "32032": {
285
+ "content": "<extra_id_67>",
286
+ "lstrip": false,
287
+ "normalized": false,
288
+ "rstrip": false,
289
+ "single_word": false,
290
+ "special": true
291
+ },
292
+ "32033": {
293
+ "content": "<extra_id_66>",
294
+ "lstrip": false,
295
+ "normalized": false,
296
+ "rstrip": false,
297
+ "single_word": false,
298
+ "special": true
299
+ },
300
+ "32034": {
301
+ "content": "<extra_id_65>",
302
+ "lstrip": false,
303
+ "normalized": false,
304
+ "rstrip": false,
305
+ "single_word": false,
306
+ "special": true
307
+ },
308
+ "32035": {
309
+ "content": "<extra_id_64>",
310
+ "lstrip": false,
311
+ "normalized": false,
312
+ "rstrip": false,
313
+ "single_word": false,
314
+ "special": true
315
+ },
316
+ "32036": {
317
+ "content": "<extra_id_63>",
318
+ "lstrip": false,
319
+ "normalized": false,
320
+ "rstrip": false,
321
+ "single_word": false,
322
+ "special": true
323
+ },
324
+ "32037": {
325
+ "content": "<extra_id_62>",
326
+ "lstrip": false,
327
+ "normalized": false,
328
+ "rstrip": false,
329
+ "single_word": false,
330
+ "special": true
331
+ },
332
+ "32038": {
333
+ "content": "<extra_id_61>",
334
+ "lstrip": false,
335
+ "normalized": false,
336
+ "rstrip": false,
337
+ "single_word": false,
338
+ "special": true
339
+ },
340
+ "32039": {
341
+ "content": "<extra_id_60>",
342
+ "lstrip": false,
343
+ "normalized": false,
344
+ "rstrip": false,
345
+ "single_word": false,
346
+ "special": true
347
+ },
348
+ "32040": {
349
+ "content": "<extra_id_59>",
350
+ "lstrip": false,
351
+ "normalized": false,
352
+ "rstrip": false,
353
+ "single_word": false,
354
+ "special": true
355
+ },
356
+ "32041": {
357
+ "content": "<extra_id_58>",
358
+ "lstrip": false,
359
+ "normalized": false,
360
+ "rstrip": false,
361
+ "single_word": false,
362
+ "special": true
363
+ },
364
+ "32042": {
365
+ "content": "<extra_id_57>",
366
+ "lstrip": false,
367
+ "normalized": false,
368
+ "rstrip": false,
369
+ "single_word": false,
370
+ "special": true
371
+ },
372
+ "32043": {
373
+ "content": "<extra_id_56>",
374
+ "lstrip": false,
375
+ "normalized": false,
376
+ "rstrip": false,
377
+ "single_word": false,
378
+ "special": true
379
+ },
380
+ "32044": {
381
+ "content": "<extra_id_55>",
382
+ "lstrip": false,
383
+ "normalized": false,
384
+ "rstrip": false,
385
+ "single_word": false,
386
+ "special": true
387
+ },
388
+ "32045": {
389
+ "content": "<extra_id_54>",
390
+ "lstrip": false,
391
+ "normalized": false,
392
+ "rstrip": false,
393
+ "single_word": false,
394
+ "special": true
395
+ },
396
+ "32046": {
397
+ "content": "<extra_id_53>",
398
+ "lstrip": false,
399
+ "normalized": false,
400
+ "rstrip": false,
401
+ "single_word": false,
402
+ "special": true
403
+ },
404
+ "32047": {
405
+ "content": "<extra_id_52>",
406
+ "lstrip": false,
407
+ "normalized": false,
408
+ "rstrip": false,
409
+ "single_word": false,
410
+ "special": true
411
+ },
412
+ "32048": {
413
+ "content": "<extra_id_51>",
414
+ "lstrip": false,
415
+ "normalized": false,
416
+ "rstrip": false,
417
+ "single_word": false,
418
+ "special": true
419
+ },
420
+ "32049": {
421
+ "content": "<extra_id_50>",
422
+ "lstrip": false,
423
+ "normalized": false,
424
+ "rstrip": false,
425
+ "single_word": false,
426
+ "special": true
427
+ },
428
+ "32050": {
429
+ "content": "<extra_id_49>",
430
+ "lstrip": false,
431
+ "normalized": false,
432
+ "rstrip": false,
433
+ "single_word": false,
434
+ "special": true
435
+ },
436
+ "32051": {
437
+ "content": "<extra_id_48>",
438
+ "lstrip": false,
439
+ "normalized": false,
440
+ "rstrip": false,
441
+ "single_word": false,
442
+ "special": true
443
+ },
444
+ "32052": {
445
+ "content": "<extra_id_47>",
446
+ "lstrip": false,
447
+ "normalized": false,
448
+ "rstrip": false,
449
+ "single_word": false,
450
+ "special": true
451
+ },
452
+ "32053": {
453
+ "content": "<extra_id_46>",
454
+ "lstrip": false,
455
+ "normalized": false,
456
+ "rstrip": false,
457
+ "single_word": false,
458
+ "special": true
459
+ },
460
+ "32054": {
461
+ "content": "<extra_id_45>",
462
+ "lstrip": false,
463
+ "normalized": false,
464
+ "rstrip": false,
465
+ "single_word": false,
466
+ "special": true
467
+ },
468
+ "32055": {
469
+ "content": "<extra_id_44>",
470
+ "lstrip": false,
471
+ "normalized": false,
472
+ "rstrip": false,
473
+ "single_word": false,
474
+ "special": true
475
+ },
476
+ "32056": {
477
+ "content": "<extra_id_43>",
478
+ "lstrip": false,
479
+ "normalized": false,
480
+ "rstrip": false,
481
+ "single_word": false,
482
+ "special": true
483
+ },
484
+ "32057": {
485
+ "content": "<extra_id_42>",
486
+ "lstrip": false,
487
+ "normalized": false,
488
+ "rstrip": false,
489
+ "single_word": false,
490
+ "special": true
491
+ },
492
+ "32058": {
493
+ "content": "<extra_id_41>",
494
+ "lstrip": false,
495
+ "normalized": false,
496
+ "rstrip": false,
497
+ "single_word": false,
498
+ "special": true
499
+ },
500
+ "32059": {
501
+ "content": "<extra_id_40>",
502
+ "lstrip": false,
503
+ "normalized": false,
504
+ "rstrip": false,
505
+ "single_word": false,
506
+ "special": true
507
+ },
508
+ "32060": {
509
+ "content": "<extra_id_39>",
510
+ "lstrip": false,
511
+ "normalized": false,
512
+ "rstrip": false,
513
+ "single_word": false,
514
+ "special": true
515
+ },
516
+ "32061": {
517
+ "content": "<extra_id_38>",
518
+ "lstrip": false,
519
+ "normalized": false,
520
+ "rstrip": false,
521
+ "single_word": false,
522
+ "special": true
523
+ },
524
+ "32062": {
525
+ "content": "<extra_id_37>",
526
+ "lstrip": false,
527
+ "normalized": false,
528
+ "rstrip": false,
529
+ "single_word": false,
530
+ "special": true
531
+ },
532
+ "32063": {
533
+ "content": "<extra_id_36>",
534
+ "lstrip": false,
535
+ "normalized": false,
536
+ "rstrip": false,
537
+ "single_word": false,
538
+ "special": true
539
+ },
540
+ "32064": {
541
+ "content": "<extra_id_35>",
542
+ "lstrip": false,
543
+ "normalized": false,
544
+ "rstrip": false,
545
+ "single_word": false,
546
+ "special": true
547
+ },
548
+ "32065": {
549
+ "content": "<extra_id_34>",
550
+ "lstrip": false,
551
+ "normalized": false,
552
+ "rstrip": false,
553
+ "single_word": false,
554
+ "special": true
555
+ },
556
+ "32066": {
557
+ "content": "<extra_id_33>",
558
+ "lstrip": false,
559
+ "normalized": false,
560
+ "rstrip": false,
561
+ "single_word": false,
562
+ "special": true
563
+ },
564
+ "32067": {
565
+ "content": "<extra_id_32>",
566
+ "lstrip": false,
567
+ "normalized": false,
568
+ "rstrip": false,
569
+ "single_word": false,
570
+ "special": true
571
+ },
572
+ "32068": {
573
+ "content": "<extra_id_31>",
574
+ "lstrip": false,
575
+ "normalized": false,
576
+ "rstrip": false,
577
+ "single_word": false,
578
+ "special": true
579
+ },
580
+ "32069": {
581
+ "content": "<extra_id_30>",
582
+ "lstrip": false,
583
+ "normalized": false,
584
+ "rstrip": false,
585
+ "single_word": false,
586
+ "special": true
587
+ },
588
+ "32070": {
589
+ "content": "<extra_id_29>",
590
+ "lstrip": false,
591
+ "normalized": false,
592
+ "rstrip": false,
593
+ "single_word": false,
594
+ "special": true
595
+ },
596
+ "32071": {
597
+ "content": "<extra_id_28>",
598
+ "lstrip": false,
599
+ "normalized": false,
600
+ "rstrip": false,
601
+ "single_word": false,
602
+ "special": true
603
+ },
604
+ "32072": {
605
+ "content": "<extra_id_27>",
606
+ "lstrip": false,
607
+ "normalized": false,
608
+ "rstrip": false,
609
+ "single_word": false,
610
+ "special": true
611
+ },
612
+ "32073": {
613
+ "content": "<extra_id_26>",
614
+ "lstrip": false,
615
+ "normalized": false,
616
+ "rstrip": false,
617
+ "single_word": false,
618
+ "special": true
619
+ },
620
+ "32074": {
621
+ "content": "<extra_id_25>",
622
+ "lstrip": false,
623
+ "normalized": false,
624
+ "rstrip": false,
625
+ "single_word": false,
626
+ "special": true
627
+ },
628
+ "32075": {
629
+ "content": "<extra_id_24>",
630
+ "lstrip": false,
631
+ "normalized": false,
632
+ "rstrip": false,
633
+ "single_word": false,
634
+ "special": true
635
+ },
636
+ "32076": {
637
+ "content": "<extra_id_23>",
638
+ "lstrip": false,
639
+ "normalized": false,
640
+ "rstrip": false,
641
+ "single_word": false,
642
+ "special": true
643
+ },
644
+ "32077": {
645
+ "content": "<extra_id_22>",
646
+ "lstrip": false,
647
+ "normalized": false,
648
+ "rstrip": false,
649
+ "single_word": false,
650
+ "special": true
651
+ },
652
+ "32078": {
653
+ "content": "<extra_id_21>",
654
+ "lstrip": false,
655
+ "normalized": false,
656
+ "rstrip": false,
657
+ "single_word": false,
658
+ "special": true
659
+ },
660
+ "32079": {
661
+ "content": "<extra_id_20>",
662
+ "lstrip": false,
663
+ "normalized": false,
664
+ "rstrip": false,
665
+ "single_word": false,
666
+ "special": true
667
+ },
668
+ "32080": {
669
+ "content": "<extra_id_19>",
670
+ "lstrip": false,
671
+ "normalized": false,
672
+ "rstrip": false,
673
+ "single_word": false,
674
+ "special": true
675
+ },
676
+ "32081": {
677
+ "content": "<extra_id_18>",
678
+ "lstrip": false,
679
+ "normalized": false,
680
+ "rstrip": false,
681
+ "single_word": false,
682
+ "special": true
683
+ },
684
+ "32082": {
685
+ "content": "<extra_id_17>",
686
+ "lstrip": false,
687
+ "normalized": false,
688
+ "rstrip": false,
689
+ "single_word": false,
690
+ "special": true
691
+ },
692
+ "32083": {
693
+ "content": "<extra_id_16>",
694
+ "lstrip": false,
695
+ "normalized": false,
696
+ "rstrip": false,
697
+ "single_word": false,
698
+ "special": true
699
+ },
700
+ "32084": {
701
+ "content": "<extra_id_15>",
702
+ "lstrip": false,
703
+ "normalized": false,
704
+ "rstrip": false,
705
+ "single_word": false,
706
+ "special": true
707
+ },
708
+ "32085": {
709
+ "content": "<extra_id_14>",
710
+ "lstrip": false,
711
+ "normalized": false,
712
+ "rstrip": false,
713
+ "single_word": false,
714
+ "special": true
715
+ },
716
+ "32086": {
717
+ "content": "<extra_id_13>",
718
+ "lstrip": false,
719
+ "normalized": false,
720
+ "rstrip": false,
721
+ "single_word": false,
722
+ "special": true
723
+ },
724
+ "32087": {
725
+ "content": "<extra_id_12>",
726
+ "lstrip": false,
727
+ "normalized": false,
728
+ "rstrip": false,
729
+ "single_word": false,
730
+ "special": true
731
+ },
732
+ "32088": {
733
+ "content": "<extra_id_11>",
734
+ "lstrip": false,
735
+ "normalized": false,
736
+ "rstrip": false,
737
+ "single_word": false,
738
+ "special": true
739
+ },
740
+ "32089": {
741
+ "content": "<extra_id_10>",
742
+ "lstrip": false,
743
+ "normalized": false,
744
+ "rstrip": false,
745
+ "single_word": false,
746
+ "special": true
747
+ },
748
+ "32090": {
749
+ "content": "<extra_id_9>",
750
+ "lstrip": false,
751
+ "normalized": false,
752
+ "rstrip": false,
753
+ "single_word": false,
754
+ "special": true
755
+ },
756
+ "32091": {
757
+ "content": "<extra_id_8>",
758
+ "lstrip": false,
759
+ "normalized": false,
760
+ "rstrip": false,
761
+ "single_word": false,
762
+ "special": true
763
+ },
764
+ "32092": {
765
+ "content": "<extra_id_7>",
766
+ "lstrip": false,
767
+ "normalized": false,
768
+ "rstrip": false,
769
+ "single_word": false,
770
+ "special": true
771
+ },
772
+ "32093": {
773
+ "content": "<extra_id_6>",
774
+ "lstrip": false,
775
+ "normalized": false,
776
+ "rstrip": false,
777
+ "single_word": false,
778
+ "special": true
779
+ },
780
+ "32094": {
781
+ "content": "<extra_id_5>",
782
+ "lstrip": false,
783
+ "normalized": false,
784
+ "rstrip": false,
785
+ "single_word": false,
786
+ "special": true
787
+ },
788
+ "32095": {
789
+ "content": "<extra_id_4>",
790
+ "lstrip": false,
791
+ "normalized": false,
792
+ "rstrip": false,
793
+ "single_word": false,
794
+ "special": true
795
+ },
796
+ "32096": {
797
+ "content": "<extra_id_3>",
798
+ "lstrip": false,
799
+ "normalized": false,
800
+ "rstrip": false,
801
+ "single_word": false,
802
+ "special": true
803
+ },
804
+ "32097": {
805
+ "content": "<extra_id_2>",
806
+ "lstrip": false,
807
+ "normalized": false,
808
+ "rstrip": false,
809
+ "single_word": false,
810
+ "special": true
811
+ },
812
+ "32098": {
813
+ "content": "<extra_id_1>",
814
+ "lstrip": false,
815
+ "normalized": false,
816
+ "rstrip": false,
817
+ "single_word": false,
818
+ "special": true
819
+ },
820
+ "32099": {
821
+ "content": "<extra_id_0>",
822
+ "lstrip": false,
823
+ "normalized": false,
824
+ "rstrip": false,
825
+ "single_word": false,
826
+ "special": true
827
+ }
828
+ },
829
+ "additional_special_tokens": [
830
+ "<extra_id_0>",
831
+ "<extra_id_1>",
832
+ "<extra_id_2>",
833
+ "<extra_id_3>",
834
+ "<extra_id_4>",
835
+ "<extra_id_5>",
836
+ "<extra_id_6>",
837
+ "<extra_id_7>",
838
+ "<extra_id_8>",
839
+ "<extra_id_9>",
840
+ "<extra_id_10>",
841
+ "<extra_id_11>",
842
+ "<extra_id_12>",
843
+ "<extra_id_13>",
844
+ "<extra_id_14>",
845
+ "<extra_id_15>",
846
+ "<extra_id_16>",
847
+ "<extra_id_17>",
848
+ "<extra_id_18>",
849
+ "<extra_id_19>",
850
+ "<extra_id_20>",
851
+ "<extra_id_21>",
852
+ "<extra_id_22>",
853
+ "<extra_id_23>",
854
+ "<extra_id_24>",
855
+ "<extra_id_25>",
856
+ "<extra_id_26>",
857
+ "<extra_id_27>",
858
+ "<extra_id_28>",
859
+ "<extra_id_29>",
860
+ "<extra_id_30>",
861
+ "<extra_id_31>",
862
+ "<extra_id_32>",
863
+ "<extra_id_33>",
864
+ "<extra_id_34>",
865
+ "<extra_id_35>",
866
+ "<extra_id_36>",
867
+ "<extra_id_37>",
868
+ "<extra_id_38>",
869
+ "<extra_id_39>",
870
+ "<extra_id_40>",
871
+ "<extra_id_41>",
872
+ "<extra_id_42>",
873
+ "<extra_id_43>",
874
+ "<extra_id_44>",
875
+ "<extra_id_45>",
876
+ "<extra_id_46>",
877
+ "<extra_id_47>",
878
+ "<extra_id_48>",
879
+ "<extra_id_49>",
880
+ "<extra_id_50>",
881
+ "<extra_id_51>",
882
+ "<extra_id_52>",
883
+ "<extra_id_53>",
884
+ "<extra_id_54>",
885
+ "<extra_id_55>",
886
+ "<extra_id_56>",
887
+ "<extra_id_57>",
888
+ "<extra_id_58>",
889
+ "<extra_id_59>",
890
+ "<extra_id_60>",
891
+ "<extra_id_61>",
892
+ "<extra_id_62>",
893
+ "<extra_id_63>",
894
+ "<extra_id_64>",
895
+ "<extra_id_65>",
896
+ "<extra_id_66>",
897
+ "<extra_id_67>",
898
+ "<extra_id_68>",
899
+ "<extra_id_69>",
900
+ "<extra_id_70>",
901
+ "<extra_id_71>",
902
+ "<extra_id_72>",
903
+ "<extra_id_73>",
904
+ "<extra_id_74>",
905
+ "<extra_id_75>",
906
+ "<extra_id_76>",
907
+ "<extra_id_77>",
908
+ "<extra_id_78>",
909
+ "<extra_id_79>",
910
+ "<extra_id_80>",
911
+ "<extra_id_81>",
912
+ "<extra_id_82>",
913
+ "<extra_id_83>",
914
+ "<extra_id_84>",
915
+ "<extra_id_85>",
916
+ "<extra_id_86>",
917
+ "<extra_id_87>",
918
+ "<extra_id_88>",
919
+ "<extra_id_89>",
920
+ "<extra_id_90>",
921
+ "<extra_id_91>",
922
+ "<extra_id_92>",
923
+ "<extra_id_93>",
924
+ "<extra_id_94>",
925
+ "<extra_id_95>",
926
+ "<extra_id_96>",
927
+ "<extra_id_97>",
928
+ "<extra_id_98>",
929
+ "<extra_id_99>"
930
+ ],
931
+ "clean_up_tokenization_spaces": true,
932
+ "eos_token": "</s>",
933
+ "extra_ids": 100,
934
+ "legacy": true,
935
+ "model_max_length": 512,
936
+ "pad_token": "<pad>",
937
+ "sp_model_kwargs": {},
938
+ "tokenizer_class": "T5Tokenizer",
939
+ "unk_token": "<unk>"
940
+ }
flux/transformer/config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_class_name": "FluxTransformer2DModel",
3
+ "_diffusers_version": "0.30.0.dev0",
4
+ "attention_head_dim": 128,
5
+ "guidance_embeds": false,
6
+ "in_channels": 64,
7
+ "joint_attention_dim": 4096,
8
+ "num_attention_heads": 24,
9
+ "num_layers": 19,
10
+ "num_single_layers": 38,
11
+ "patch_size": 1,
12
+ "pooled_projection_dim": 768
13
+ }
flux/transformer/diffusion_pytorch_model-00001-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9b633dbe87316385c5b1c262bd4b5a01e3d955170661d63dcec8a01e89c0d820
3
+ size 9962580296
flux/transformer/diffusion_pytorch_model-00002-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:58b4434078f0c2567ddc54e3b5cbf39626ab55fbd9d5c22956e183668f535dec
3
+ size 9949328904
flux/transformer/diffusion_pytorch_model-00003-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e2cbc25471ed5186e69a9b51098300cb2f612556453e38a372c851a220ed238d
3
+ size 3870584832
flux/transformer/diffusion_pytorch_model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
flux/vae/config.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_class_name": "AutoencoderKL",
3
+ "_diffusers_version": "0.30.0.dev0",
4
+ "act_fn": "silu",
5
+ "block_out_channels": [
6
+ 128,
7
+ 256,
8
+ 512,
9
+ 512
10
+ ],
11
+ "down_block_types": [
12
+ "DownEncoderBlock2D",
13
+ "DownEncoderBlock2D",
14
+ "DownEncoderBlock2D",
15
+ "DownEncoderBlock2D"
16
+ ],
17
+ "force_upcast": true,
18
+ "in_channels": 3,
19
+ "latent_channels": 16,
20
+ "latents_mean": null,
21
+ "latents_std": null,
22
+ "layers_per_block": 2,
23
+ "mid_block_add_attention": true,
24
+ "norm_num_groups": 32,
25
+ "out_channels": 3,
26
+ "sample_size": 1024,
27
+ "scaling_factor": 0.3611,
28
+ "shift_factor": 0.1159,
29
+ "up_block_types": [
30
+ "UpDecoderBlock2D",
31
+ "UpDecoderBlock2D",
32
+ "UpDecoderBlock2D",
33
+ "UpDecoderBlock2D"
34
+ ],
35
+ "use_post_quant_conv": false,
36
+ "use_quant_conv": false
37
+ }
flux/vae/diffusion_pytorch_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5b59a26851551b67ae1fe58d32e76486e1e812def4696a4bea97f16604d40a3
3
+ size 167666902
qwen3-tts/.gitattributes ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
qwen3-tts/README.md ADDED
@@ -0,0 +1,1315 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ ---
4
+
5
+ # Qwen3-TTS
6
+
7
+ ## Overview
8
+ ### Introduction
9
+
10
+ <p align="center">
11
+ <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-TTS-Repo/qwen3_tts_introduction.png" width="90%"/>
12
+ <p>
13
+
14
+ Qwen3-TTS covers 10 major languages (Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish, and Italian) as well as multiple dialectal voice profiles to meet global application needs. In addition, the models feature strong contextual understanding, enabling adaptive control of tone, speaking rate, and emotional expression based on instructions and text semantics, and they show markedly improved robustness to noisy input text. Key features:
15
+
16
+ * **Powerful Speech Representation**: Powered by the self-developed Qwen3-TTS-Tokenizer-12Hz, it achieves efficient acoustic compression and high-dimensional semantic modeling of speech signals. It fully preserves paralinguistic information and acoustic environmental features, enabling high-speed, high-fidelity speech reconstruction through a lightweight non-DiT architecture.
17
+ * **Universal End-to-End Architecture**: Utilizing a discrete multi-codebook LM architecture, it realizes full-information end-to-end speech modeling. This completely bypasses the information bottlenecks and cascading errors inherent in traditional LM+DiT schemes, significantly enhancing the model’s versatility, generation efficiency, and performance ceiling.
18
+ * **Extreme Low-Latency Streaming Generation**: Based on the innovative Dual-Track hybrid streaming generation architecture, a single model supports both streaming and non-streaming generation. It can output the first audio packet immediately after a single character is input, with end-to-end synthesis latency as low as 97ms, meeting the rigorous demands of real-time interactive scenarios.
19
+ * **Intelligent Text Understanding and Voice Control**: Supports speech generation driven by natural language instructions, allowing for flexible control over multi-dimensional acoustic attributes such as timbre, emotion, and prosody. By deeply integrating text semantic understanding, the model adaptively adjusts tone, rhythm, and emotional expression, achieving lifelike “what you imagine is what you hear” output.
20
+
21
+
22
+ ### Model Architecture
23
+
24
+ <p align="center">
25
+ <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-TTS-Repo/overview.png" width="80%"/>
26
+ <p>
27
+
28
+ ### Released Models Description and Download
29
+
30
+ Below is an introduction and download information for the Qwen3-TTS models that have already been released. Other models mentioned in the technical report will be released in the near future. Please select and download the model that fits your needs.
31
+
32
+ | Tokenizer Name | Description |
33
+ |---------------------------------|-------------|
34
+ | Qwen3-TTS-Tokenizer-12Hz | The Qwen3-TTS-Tokenizer-12Hz model which can encode the input speech into codes and decode them back into speech. |
35
+
36
+
37
+ | Model | Features | Language Support | Streaming | Instruction Control |
38
+ |---|---|---|---|---|
39
+ | Qwen3-TTS-12Hz-1.7B-VoiceDesign | Performs voice design based on user-provided descriptions. | Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish, Italian | ✅ | ✅ |
40
+ | Qwen3-TTS-12Hz-1.7B-CustomVoice | Provides style control over target timbres via user instructions; supports 9 premium timbres covering various combinations of gender, age, language, and dialect. | Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish, Italian | ✅ | ✅ |
41
+ | Qwen3-TTS-12Hz-1.7B-Base | Base model capable of 3-second rapid voice clone from user audio input; can be used for fine-tuning (FT) other models. | Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish, Italian | ✅ | |
42
+ | Qwen3-TTS-12Hz-0.6B-CustomVoice | Supports 9 premium timbres covering various combinations of gender, age, language, and dialect. | Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish, Italian | ✅ | |
43
+ | Qwen3-TTS-12Hz-0.6B-Base | Base model capable of 3-second rapid voice clone from user audio input; can be used for fine-tuning (FT) other models. | Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish, Italian | ✅ | |
44
+
45
+ During model loading in the qwen-tts package or vLLM, model weights will be automatically downloaded based on the model name. However, if your runtime environment is not conducive to downloading weights during execution, you can refer to the following commands to manually download the model weights to a local directory:
46
+
47
+ ```bash
48
+ # Download through ModelScope (recommended for users in Mainland China)
49
+ pip install -U modelscope
50
+ modelscope download --model Qwen/Qwen3-TTS-Tokenizer-12Hz --local_dir ./Qwen3-TTS-Tokenizer-12Hz
51
+ modelscope download --model Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice --local_dir ./Qwen3-TTS-12Hz-1.7B-CustomVoice
52
+ modelscope download --model Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign --local_dir ./Qwen3-TTS-12Hz-1.7B-VoiceDesign
53
+ modelscope download --model Qwen/Qwen3-TTS-12Hz-1.7B-Base --local_dir ./Qwen3-TTS-12Hz-1.7B-Base
54
+ modelscope download --model Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice --local_dir ./Qwen3-TTS-12Hz-0.6B-CustomVoice
55
+ modelscope download --model Qwen/Qwen3-TTS-12Hz-0.6B-Base --local_dir ./Qwen3-TTS-12Hz-0.6B-Base
56
+
57
+ # Download through Hugging Face
58
+ pip install -U "huggingface_hub[cli]"
59
+ huggingface-cli download Qwen/Qwen3-TTS-Tokenizer-12Hz --local-dir ./Qwen3-TTS-Tokenizer-12Hz
60
+ huggingface-cli download Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice --local-dir ./Qwen3-TTS-12Hz-1.7B-CustomVoice
61
+ huggingface-cli download Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign --local-dir ./Qwen3-TTS-12Hz-1.7B-VoiceDesign
62
+ huggingface-cli download Qwen/Qwen3-TTS-12Hz-1.7B-Base --local-dir ./Qwen3-TTS-12Hz-1.7B-Base
63
+ huggingface-cli download Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice --local-dir ./Qwen3-TTS-12Hz-0.6B-CustomVoice
64
+ huggingface-cli download Qwen/Qwen3-TTS-12Hz-0.6B-Base --local-dir ./Qwen3-TTS-12Hz-0.6B-Base
65
+ ```
66
+
67
+
68
+ ## Quickstart
69
+
70
+ ### Environment Setup
71
+
72
+ The easiest way to quickly use Qwen3-TTS is to install the `qwen-tts` Python package from PyPI. This will pull in the required runtime dependencies and allow you to load any released Qwen3-TTS model. We recommend using a **fresh, isolated environment** to avoid dependency conflicts with existing packages. You can create a clean Python 3.12 environment like this:
73
+
74
+ ```bash
75
+ conda create -n qwen3-tts python=3.12 -y
76
+ conda activate qwen3-tts
77
+ ```
78
+
79
+ then run:
80
+
81
+ ```bash
82
+ pip install -U qwen-tts
83
+ ```
84
+
85
+ If you want to develop or modify the code locally, install from source in editable mode.
86
+
87
+ ```bash
88
+ git clone https://github.com/QwenLM/Qwen3-TTS.git
89
+ cd Qwen3-TTS
90
+ pip install -e .
91
+ ```
92
+
93
+ Additionally, we recommend using FlashAttention 2 to reduce GPU memory usage.
94
+
95
+ ```bash
96
+ pip install -U flash-attn --no-build-isolation
97
+ ```
98
+
99
+ If your machine has less than 96GB of RAM and lots of CPU cores, run:
100
+
101
+ ```bash
102
+ MAX_JOBS=4 pip install -U flash-attn --no-build-isolation
103
+ ```
104
+
105
+ Also, you should have hardware that is compatible with FlashAttention 2. Read more about it in the official documentation of the [FlashAttention repository](https://github.com/Dao-AILab/flash-attention). FlashAttention 2 can only be used when a model is loaded in `torch.float16` or `torch.bfloat16`.
106
+
107
+
108
+ ### Python Package Usage
109
+
110
+ After installation, you can import `Qwen3TTSModel` to run custom voice TTS, voice design, and voice clone. The model weights can be specified either as a Hugging Face model id (recommended) or as a local directory path you downloaded. For all the `generate_*` functions below, besides the parameters shown and explicitly documented, you can also pass generation kwargs supported by Hugging Face Transformers `model.generate`, e.g., `max_new_tokens`, `top_p`, etc.
111
+
112
+ #### Custom Voice Generate
113
+
114
+ For custom voice models (`Qwen3-TTS-12Hz-1.7B/0.6B-CustomVoice`), you just need to call `generate_custom_voice`, passing a single string or a batch list, along with `language`, `speaker`, and optional `instruct`. You can also call `model.get_supported_speakers()` and `model.get_supported_languages()` to see which speakers and languages the current model supports.
115
+
116
+ ```python
117
+ import torch
118
+ import soundfile as sf
119
+ from qwen_tts import Qwen3TTSModel
120
+
121
+ model = Qwen3TTSModel.from_pretrained(
122
+ "Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice",
123
+ device_map="cuda:0",
124
+ dtype=torch.bfloat16,
125
+ attn_implementation="flash_attention_2",
126
+ )
127
+
128
+ # single inference
129
+ wavs, sr = model.generate_custom_voice(
130
+ text="其实我真的有发现,我是一个特别善于观察别人情绪的人。",
131
+ language="Chinese", # Pass `Auto` (or omit) for auto language adaptive; if the target language is known, set it explicitly.
132
+ speaker="Vivian",
133
+ instruct="用特别愤怒的语气说", # Omit if not needed.
134
+ )
135
+ sf.write("output_custom_voice.wav", wavs[0], sr)
136
+
137
+ # batch inference
138
+ wavs, sr = model.generate_custom_voice(
139
+ text=[
140
+ "其实我真的有发现,我是一个特别善于观察别人情绪的人。",
141
+ "She said she would be here by noon."
142
+ ],
143
+ language=["Chinese", "English"],
144
+ speaker=["Vivian", "Ryan"],
145
+ instruct=["", "Very happy."]
146
+ )
147
+ sf.write("output_custom_voice_1.wav", wavs[0], sr)
148
+ sf.write("output_custom_voice_2.wav", wavs[1], sr)
149
+ ```
150
+
151
+ For `Qwen3-TTS-12Hz-1.7B/0.6B-CustomVoice` models, the supported speaker list and speaker descriptions are provided below. We recommend using each speaker’s native language for the best quality. Of course, each speaker can speak any language supported by the model.
152
+
153
+ | Speaker | Voice Description | Native language |
154
+ | --- | --- | --- |
155
+ | Vivian | Bright, slightly edgy young female voice. | Chinese |
156
+ | Serena | Warm, gentle young female voice. | Chinese |
157
+ | Uncle_Fu | Seasoned male voice with a low, mellow timbre. | Chinese |
158
+ | Dylan | Youthful Beijing male voice with a clear, natural timbre. | Chinese (Beijing Dialect) |
159
+ | Eric | Lively Chengdu male voice with a slightly husky brightness. | Chinese (Sichuan Dialect) |
160
+ | Ryan | Dynamic male voice with strong rhythmic drive. | English |
161
+ | Aiden | Sunny American male voice with a clear midrange. | English |
162
+ | Ono_Anna | Playful Japanese female voice with a light, nimble timbre. | Japanese |
163
+ | Sohee | Warm Korean female voice with rich emotion. | Korean |
164
+
165
+ #### Voice Design
166
+
167
+ For the voice design model (`Qwen3-TTS-12Hz-1.7B-VoiceDesign`), you can use `generate_voice_design` to provide the target text and a natural-language `instruct` description.
168
+
169
+ ```python
170
+ import torch
171
+ import soundfile as sf
172
+ from qwen_tts import Qwen3TTSModel
173
+
174
+ model = Qwen3TTSModel.from_pretrained(
175
+ "Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign",
176
+ device_map="cuda:0",
177
+ dtype=torch.bfloat16,
178
+ attn_implementation="flash_attention_2",
179
+ )
180
+
181
+ # single inference
182
+ wavs, sr = model.generate_voice_design(
183
+ text="哥哥,你回来啦,人家等了你好久好久了,要抱抱!",
184
+ language="Chinese",
185
+ instruct="体现撒娇稚嫩的萝莉女声,音调偏高且起伏明显,营造出黏人、做作又刻意卖萌的听觉效果。",
186
+ )
187
+ sf.write("output_voice_design.wav", wavs[0], sr)
188
+
189
+ # batch inference
190
+ wavs, sr = model.generate_voice_design(
191
+ text=[
192
+ "哥哥,你回来啦,人家等了你好久好久了,要抱抱!",
193
+ "It's in the top drawer... wait, it's empty? No way, that's impossible! I'm sure I put it there!"
194
+ ],
195
+ language=["Chinese", "English"],
196
+ instruct=[
197
+ "体现撒娇稚嫩的萝莉女声,音调偏高且起伏明显,营造出黏人、做作又刻意卖萌的听觉效果。",
198
+ "Speak in an incredulous tone, but with a hint of panic beginning to creep into your voice."
199
+ ]
200
+ )
201
+ sf.write("output_voice_design_1.wav", wavs[0], sr)
202
+ sf.write("output_voice_design_2.wav", wavs[1], sr)
203
+ ```
204
+
205
+ #### Voice Clone
206
+
207
+ For the voice clone model (`Qwen3-TTS-12Hz-1.7B/0.6B-Base`), to clone a voice and synthesize new content, you just need to provide a reference audio clip (`ref_audio`) along with its transcript (`ref_text`). `ref_audio` can be a local file path, a URL, a base64 string, or a `(numpy_array, sample_rate)` tuple. If you set `x_vector_only_mode=True`, only the speaker embedding is used so `ref_text` is not required, but cloning quality may be reduced.
208
+
209
+ ```python
210
+ import torch
211
+ import soundfile as sf
212
+ from qwen_tts import Qwen3TTSModel
213
+
214
+ model = Qwen3TTSModel.from_pretrained(
215
+ "Qwen/Qwen3-TTS-12Hz-1.7B-Base",
216
+ device_map="cuda:0",
217
+ dtype=torch.bfloat16,
218
+ attn_implementation="flash_attention_2",
219
+ )
220
+
221
+ ref_audio = "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-TTS-Repo/clone.wav"
222
+ ref_text = "Okay. Yeah. I resent you. I love you. I respect you. But you know what? You blew it! And thanks to you."
223
+
224
+ wavs, sr = model.generate_voice_clone(
225
+ text="I am solving the equation: x = [-b ± √(b²-4ac)] / 2a? Nobody can — it's a disaster (◍•͈⌔•͈◍), very sad!",
226
+ language="English",
227
+ ref_audio=ref_audio,
228
+ ref_text=ref_text,
229
+ )
230
+ sf.write("output_voice_clone.wav", wavs[0], sr)
231
+ ```
232
+
233
+ If you need to reuse the same reference prompt across multiple generations (to avoid recomputing prompt features), build it once with `create_voice_clone_prompt` and pass it via `voice_clone_prompt`.
234
+
235
+ ```python
236
+ prompt_items = model.create_voice_clone_prompt(
237
+ ref_audio=ref_audio,
238
+ ref_text=ref_text,
239
+ x_vector_only_mode=False,
240
+ )
241
+ wavs, sr = model.generate_voice_clone(
242
+ text=["Sentence A.", "Sentence B."],
243
+ language=["English", "English"],
244
+ voice_clone_prompt=prompt_items,
245
+ )
246
+ sf.write("output_voice_clone_1.wav", wavs[0], sr)
247
+ sf.write("output_voice_clone_2.wav", wavs[1], sr)
248
+ ```
249
+
250
+ For more examples of reusable voice clone prompts, batch cloning, and batch inference, please refer to the [example codes](https://github.com/QwenLM/Qwen3-TTS/blob/main/examples/test_model_12hz_base.py). With those examples and the `generate_voice_clone` function description, you can explore more advanced usage patterns.
251
+
252
+ #### Voice Design then Clone
253
+
254
+ If you want a designed voice that you can reuse like a cloned speaker, a practical workflow is: (1) use the **VoiceDesign** model to synthesize a short reference clip that matches your target persona, (2) feed that clip into `create_voice_clone_prompt` to build a reusable prompt, and then (3) call `generate_voice_clone` with `voice_clone_prompt` to generate new content without re-extracting features every time. This is especially useful when you want a consistent character voice across many lines.
255
+
256
+ ```python
257
+ import torch
258
+ import soundfile as sf
259
+ from qwen_tts import Qwen3TTSModel
260
+
261
+ # create a reference audio in the target style using the VoiceDesign model
262
+ design_model = Qwen3TTSModel.from_pretrained(
263
+ "Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign",
264
+ device_map="cuda:0",
265
+ dtype=torch.bfloat16,
266
+ attn_implementation="flash_attention_2",
267
+ )
268
+
269
+ ref_text = "H-hey! You dropped your... uh... calculus notebook? I mean, I think it's yours? Maybe?"
270
+ ref_instruct = "Male, 17 years old, tenor range, gaining confidence - deeper breath support now, though vowels still tighten when nervous"
271
+ ref_wavs, sr = design_model.generate_voice_design(
272
+ text=ref_text,
273
+ language="English",
274
+ instruct=ref_instruct
275
+ )
276
+ sf.write("voice_design_reference.wav", ref_wavs[0], sr)
277
+
278
+ # build a reusable clone prompt from the voice design reference
279
+ clone_model = Qwen3TTSModel.from_pretrained(
280
+ "Qwen/Qwen3-TTS-12Hz-1.7B-Base",
281
+ device_map="cuda:0",
282
+ dtype=torch.bfloat16,
283
+ attn_implementation="flash_attention_2",
284
+ )
285
+
286
+ voice_clone_prompt = clone_model.create_voice_clone_prompt(
287
+ ref_audio=(ref_wavs[0], sr), # or "voice_design_reference.wav"
288
+ ref_text=ref_text,
289
+ )
290
+
291
+ sentences = [
292
+ "No problem! I actually... kinda finished those already? If you want to compare answers or something...",
293
+ "What? No! I mean yes but not like... I just think you're... your titration technique is really precise!",
294
+ ]
295
+
296
+ # reuse it for multiple single calls
297
+ wavs, sr = clone_model.generate_voice_clone(
298
+ text=sentences[0],
299
+ language="English",
300
+ voice_clone_prompt=voice_clone_prompt,
301
+ )
302
+ sf.write("clone_single_1.wav", wavs[0], sr)
303
+
304
+ wavs, sr = clone_model.generate_voice_clone(
305
+ text=sentences[1],
306
+ language="English",
307
+ voice_clone_prompt=voice_clone_prompt,
308
+ )
309
+ sf.write("clone_single_2.wav", wavs[0], sr)
310
+
311
+ # or batch generate in one call
312
+ wavs, sr = clone_model.generate_voice_clone(
313
+ text=sentences,
314
+ language=["English", "English"],
315
+ voice_clone_prompt=voice_clone_prompt,
316
+ )
317
+ for i, w in enumerate(wavs):
318
+ sf.write(f"clone_batch_{i}.wav", w, sr)
319
+ ```
320
+
321
+ #### Tokenizer Encode and Decode
322
+
323
+ If you only want to encode and decode audio for transport or training and so on, `Qwen3TTSTokenizer` supports encode/decode with paths, URLs, numpy waveforms, and dict/list payloads, for example:
324
+
325
+ ```python
326
+ import soundfile as sf
327
+ from qwen_tts import Qwen3TTSTokenizer
328
+
329
+ tokenizer = Qwen3TTSTokenizer.from_pretrained(
330
+ "Qwen/Qwen3-TTS-Tokenizer-12Hz",
331
+ device_map="cuda:0",
332
+ )
333
+
334
+ enc = tokenizer.encode("https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-TTS-Repo/tokenizer_demo_1.wav")
335
+ wavs, sr = tokenizer.decode(enc)
336
+ sf.write("decode_output.wav", wavs[0], sr)
337
+ ```
338
+
339
+ For more tokenizer examples (including different input formats and batch usage), please refer to the [example codes](https://github.com/QwenLM/Qwen3-TTS/blob/main/examples/test_tokenizer_12hz.py). With those examples and the description for `Qwen3TTSTokenizer`, you can explore more advanced usage patterns.
340
+
341
+ ### Launch Local Web UI Demo
342
+
343
+ To launch the Qwen3-TTS web ui demo, simply install the `qwen-tts` package and run `qwen-tts-demo`. Use the command below for help:
344
+
345
+ ```bash
346
+ qwen-tts-demo --help
347
+ ```
348
+
349
+ To launch the demo, you can use the following commands:
350
+
351
+ ```bash
352
+ # CustomVoice model
353
+ qwen-tts-demo Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice --ip 0.0.0.0 --port 8000
354
+ # VoiceDesign model
355
+ qwen-tts-demo Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign --ip 0.0.0.0 --port 8000
356
+ # Base model
357
+ qwen-tts-demo Qwen/Qwen3-TTS-12Hz-1.7B-Base --ip 0.0.0.0 --port 8000
358
+ ```
359
+
360
+ And then open `http://<your-ip>:8000`, or access it via port forwarding in tools like VS Code.
361
+
362
+ #### Base Model HTTPS Notes
363
+
364
+ To avoid browser microphone permission issues after deploying the server, for Base model deployments, it is recommended/required to run the gradio service over **HTTPS** (especially when accessed remotely or behind modern browsers/gateways). Use `--ssl-certfile` and `--ssl-keyfile` to enable HTTPS. First we need to generate a private key and a self-signed cert (valid for 365 days):
365
+
366
+ ```bash
367
+ openssl req -x509 -newkey rsa:2048 \
368
+ -keyout key.pem -out cert.pem \
369
+ -days 365 -nodes \
370
+ -subj "/CN=localhost"
371
+ ```
372
+
373
+ Then run the demo with HTTPS:
374
+
375
+ ```bash
376
+ qwen-tts-demo Qwen/Qwen3-TTS-12Hz-1.7B-Base \
377
+ --ip 0.0.0.0 --port 8000 \
378
+ --ssl-certfile cert.pem \
379
+ --ssl-keyfile key.pem \
380
+ --no-ssl-verify
381
+ ```
382
+
383
+ And open `https://<your-ip>:8000` to experience it. If your browser shows a warning, it’s expected for self-signed certificates. For production, use a real certificate.
384
+
385
+ ### DashScope API Usage
386
+
387
+ To further explore Qwen3-TTS, we encourage you to try our DashScope API for a faster and more efficient experience. For detailed API information and documentation, please refer to the following:
388
+
389
+ | API Description | API Documentation (Mainland China) | API Documentation (International) |
390
+ |------------------|-----------------------------------|------------------------------------|
391
+ | Real-time API for Qwen3-TTS of custom voice model. | [https://help.aliyun.com/zh/model-studio/qwen-tts-realtime](https://help.aliyun.com/zh/model-studio/qwen-tts-realtime) | [https://www.alibabacloud.com/help/en/model-studio/qwen-tts-realtime](https://www.alibabacloud.com/help/en/model-studio/qwen-tts-realtime) |
392
+ | Real-time API for Qwen3-TTS of voice clone model. | [https://help.aliyun.com/zh/model-studio/qwen-tts-voice-cloning](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-cloning) | [https://www.alibabacloud.com/help/en/model-studio/qwen-tts-voice-cloning](https://www.alibabacloud.com/help/en/model-studio/qwen-tts-voice-cloning) |
393
+ | Real-time API for Qwen3-TTS of voice design model. | [https://help.aliyun.com/zh/model-studio/qwen-tts-voice-design](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-design) | [https://www.alibabacloud.com/help/en/model-studio/qwen-tts-voice-design](https://www.alibabacloud.com/help/en/model-studio/qwen-tts-voice-design) |
394
+
395
+
396
+ ## vLLM Usage
397
+
398
+ vLLM officially provides day-0 support for Qwen3-TTS! Welcome to use vLLM-Omni for Qwen3-TTS deployment and inference. For installation and more details, please check [vLLM-Omni official documentation](https://docs.vllm.ai/projects/vllm-omni/en/latest/getting_started/quickstart/#installation). Now only offline inference is supported. Online serving will be supported later, and vLLM-Omni will continue to offer support and optimization for Qwen3-TTS in areas such as inference speed and streaming capabilities.
399
+
400
+ ### Offline Inference
401
+ You can use vLLM-Omni to inference Qwen3-TTS locally, we provide examples in [vLLM-Omni repo](https://github.com/vllm-project/vllm-omni/tree/main/examples/offline_inference/qwen3_tts) which can generate audio output:
402
+ ```bash
403
+ # git clone https://github.com/vllm-project/vllm-omni.git
404
+
405
+ # cd vllm-omni/examples/offline_inference/qwen3_tts
406
+
407
+ # Run a single sample with CustomVoice task
408
+ python end2end.py --query-type CustomVoice
409
+
410
+ # Batch sample (multiple prompts in one run) with CustomVoice task:
411
+ python end2end.py --query-type CustomVoice --use-batch-sample
412
+
413
+ # Run a single sample with VoiceDesign task
414
+ python end2end.py --query-type VoiceDesign
415
+
416
+ # Batch sample (multiple prompts in one run) with VoiceDesign task:
417
+ python end2end.py --query-type VoiceDesign --use-batch-sample
418
+
419
+ # Run a single sample with Base task in icl mode-tag
420
+ python end2end.py --query-type Base --mode-tag icl
421
+ ```
422
+
423
+ ## Evaluation
424
+
425
+ During evaluation, we ran inference for all models with `dtype=torch.bfloat16` and set `max_new_tokens=2048`. All other sampling parameters used the defaults from the checkpoint’s `generate_config.json`. For the Seed-Test and InstructTTS-Eval test sets, we set `language="auto"`, while for all other test sets we explicitly passed the corresponding `language`. The detailed results are shown below.
426
+
427
+
428
+ <details>
429
+ <summary>Speech Generation Benchmarks</summary>
430
+
431
+ *Zero-shot speech generation on the Seed-TTS test set. Performance is measured by Word Error Rate (WER, ↓), where lower is better.*
432
+
433
+ <table>
434
+ <thead>
435
+ <tr>
436
+ <th style="text-align: center;">Datasets</th>
437
+ <th style="text-align: left;">Model</th>
438
+ <th colspan="2" style="text-align: center;">Performance</th>
439
+ </tr>
440
+ <tr style="border-bottom: 1px solid #ddd; border-top: 1px solid #ddd;">
441
+ <td colspan="4" style="text-align: center;"><em>Content Consistency</em></td>
442
+ </tr>
443
+ </thead>
444
+ <tbody>
445
+ <tr>
446
+ <td rowspan="14" style="text-align: center; vertical-align: middle;">SEED<br><em>test-zh</em> | <em>test-en</em></td>
447
+ <td style="text-align: left;">Seed-TTS (Anastassiou et al., 2024)</td>
448
+ <td style="text-align: center;">1.12</td>
449
+ <td style="text-align: center;">2.25</td>
450
+ </tr>
451
+ <tr>
452
+ <td style="text-align: left;">MaskGCT (Wang et al., 2024)</td>
453
+ <td style="text-align: center;">2.27</td>
454
+ <td style="text-align: center;">2.62</td>
455
+ </tr>
456
+ <tr>
457
+ <td style="text-align: left;">E2 TTS (Eskimez et al., 2024)</td>
458
+ <td style="text-align: center;">1.97</td>
459
+ <td style="text-align: center;">2.19</td>
460
+ </tr>
461
+ <tr>
462
+ <td style="text-align: left;">F5-TTS (Chen et al., 2024)</td>
463
+ <td style="text-align: center;">1.56</td>
464
+ <td style="text-align: center;">1.83</td>
465
+ </tr>
466
+ <tr>
467
+ <td style="text-align: left;">Spark TTS (Wang et al., 2025)</td>
468
+ <td style="text-align: center;">1.20</td>
469
+ <td style="text-align: center;">1.98</td>
470
+ </tr>
471
+ <tr>
472
+ <td style="text-align: left;">Llasa-8B (Ye et al., 2025b)</td>
473
+ <td style="text-align: center;">1.59</td>
474
+ <td style="text-align: center;">2.97</td>
475
+ </tr>
476
+ <tr>
477
+ <td style="text-align: left;">KALL-E (Xia et al., 2024)</td>
478
+ <td style="text-align: center;">0.96</td>
479
+ <td style="text-align: center;">1.94</td>
480
+ </tr>
481
+ <tr>
482
+ <td style="text-align: left;">FireRedTTS 2 (Xie et al., 2025)</td>
483
+ <td style="text-align: center;">1.14</td>
484
+ <td style="text-align: center;">1.95</td>
485
+ </tr>
486
+ <tr>
487
+ <td style="text-align: left;">CosyVoice 3 (Du et al., 2025)</td>
488
+ <td style="text-align: center;"><strong>0.71</strong></td>
489
+ <td style="text-align: center;">1.45</td>
490
+ </tr>
491
+ <tr>
492
+ <td style="text-align: left;">MiniMax-Speech (Zhang et al., 2025a)</td>
493
+ <td style="text-align: center;">0.83</td>
494
+ <td style="text-align: center;">1.65</td>
495
+ </tr>
496
+ <tr>
497
+ <td style="text-align: left;">Qwen3-TTS-25Hz-0.6B-Base</td>
498
+ <td style="text-align: center;">1.18</td>
499
+ <td style="text-align: center;">1.64</td>
500
+ </tr>
501
+ <tr>
502
+ <td style="text-align: left;">Qwen3-TTS-25Hz-1.7B-Base</td>
503
+ <td style="text-align: center;">1.10</td>
504
+ <td style="text-align: center;">1.49</td>
505
+ </tr>
506
+ <tr>
507
+ <td style="text-align: left;">Qwen3-TTS-12Hz-0.6B-Base</td>
508
+ <td style="text-align: center;">0.92</td>
509
+ <td style="text-align: center;">1.32</td>
510
+ </tr>
511
+ <tr>
512
+ <td style="text-align: left;">Qwen3-TTS-12Hz-1.7B-Base</td>
513
+ <td style="text-align: center;">0.77</td>
514
+ <td style="text-align: center;"><strong>1.24</strong></td>
515
+ </tr>
516
+ </tbody>
517
+ </table>
518
+
519
+ <br>
520
+
521
+ *Multilingual speech generation on the TTS multilingual test set. Performance is measured by Word Error Rate (WER, ↓) for content consistency and Cosine Similarity (SIM, ↑) for speaker similarity.*
522
+
523
+ <table>
524
+ <thead>
525
+ <tr>
526
+ <th rowspan="2" style="text-align: left; vertical-align: bottom;">Language</th>
527
+ <th colspan="2" style="text-align: center;">Qwen3-TTS-25Hz</th>
528
+ <th colspan="2" style="text-align: center;">Qwen3-TTS-12Hz</th>
529
+ <th rowspan="2" style="text-align: center; vertical-align: bottom;">MiniMax</th>
530
+ <th rowspan="2" style="text-align: center; vertical-align: bottom;">ElevenLabs</th>
531
+ </tr>
532
+ <tr>
533
+ <th style="text-align: center;">0.6B-Base</th>
534
+ <th style="text-align: center;">1.7B-Base</th>
535
+ <th style="text-align: center;">0.6B-Base</th>
536
+ <th style="text-align: center;">1.7B-Base</th>
537
+ </tr>
538
+ </thead>
539
+ <tbody>
540
+ <tr>
541
+ <td colspan="7" style="text-align: center; border-top: 1px solid #ddd; border-bottom: 1px solid #ddd;"><em>Content Consistency</em></td>
542
+ </tr>
543
+ <tr>
544
+ <td style="text-align: left;">Chinese</td>
545
+ <td style="text-align: center;">1.108</td>
546
+ <td style="text-align: center;"><strong>0.777</strong></td>
547
+ <td style="text-align: center;">1.145</td>
548
+ <td style="text-align: center;">0.928</td>
549
+ <td style="text-align: center;">2.252</td>
550
+ <td style="text-align: center;">16.026</td>
551
+ </tr>
552
+ <tr>
553
+ <td style="text-align: left;">English</td>
554
+ <td style="text-align: center;">1.048</td>
555
+ <td style="text-align: center;">1.014</td>
556
+ <td style="text-align: center;"><strong>0.836</strong></td>
557
+ <td style="text-align: center;">0.934</td>
558
+ <td style="text-align: center;">2.164</td>
559
+ <td style="text-align: center;">2.339</td>
560
+ </tr>
561
+ <tr>
562
+ <td style="text-align: left;">German</td>
563
+ <td style="text-align: center;">1.501</td>
564
+ <td style="text-align: center;">0.960</td>
565
+ <td style="text-align: center;">1.089</td>
566
+ <td style="text-align: center;">1.235</td>
567
+ <td style="text-align: center;">1.906</td>
568
+ <td style="text-align: center;"><strong>0.572</strong></td>
569
+ </tr>
570
+ <tr>
571
+ <td style="text-align: left;">Italian</td>
572
+ <td style="text-align: center;">1.169</td>
573
+ <td style="text-align: center;">1.105</td>
574
+ <td style="text-align: center;">1.534</td>
575
+ <td style="text-align: center;"><strong>0.948</strong></td>
576
+ <td style="text-align: center;">1.543</td>
577
+ <td style="text-align: center;">1.743</td>
578
+ </tr>
579
+ <tr>
580
+ <td style="text-align: left;">Portuguese</td>
581
+ <td style="text-align: center;">2.046</td>
582
+ <td style="text-align: center;">1.778</td>
583
+ <td style="text-align: center;">2.254</td>
584
+ <td style="text-align: center;">1.526</td>
585
+ <td style="text-align: center;">1.877</td>
586
+ <td style="text-align: center;"><strong>1.331</strong></td>
587
+ </tr>
588
+ <tr>
589
+ <td style="text-align: left;">Spanish</td>
590
+ <td style="text-align: center;">2.031</td>
591
+ <td style="text-align: center;">1.491</td>
592
+ <td style="text-align: center;">1.491</td>
593
+ <td style="text-align: center;">1.126</td>
594
+ <td style="text-align: center;"><strong>1.029</strong></td>
595
+ <td style="text-align: center;">1.084</td>
596
+ </tr>
597
+ <tr>
598
+ <td style="text-align: left;">Japanese</td>
599
+ <td style="text-align: center;">4.189</td>
600
+ <td style="text-align: center;">5.121</td>
601
+ <td style="text-align: center;">6.404</td>
602
+ <td style="text-align: center;">3.823</td>
603
+ <td style="text-align: center;"><strong>3.519</strong></td>
604
+ <td style="text-align: center;">10.646</td>
605
+ </tr>
606
+ <tr>
607
+ <td style="text-align: left;">Korean</td>
608
+ <td style="text-align: center;">2.852</td>
609
+ <td style="text-align: center;">2.631</td>
610
+ <td style="text-align: center;"><strong>1.741</strong></td>
611
+ <td style="text-align: center;">1.755</td>
612
+ <td style="text-align: center;">1.747</td>
613
+ <td style="text-align: center;">1.865</td>
614
+ </tr>
615
+ <tr>
616
+ <td style="text-align: left;">French</td>
617
+ <td style="text-align: center;">2.852</td>
618
+ <td style="text-align: center;"><strong>2.631</strong></td>
619
+ <td style="text-align: center;">2.931</td>
620
+ <td style="text-align: center;">2.858</td>
621
+ <td style="text-align: center;">4.099</td>
622
+ <td style="text-align: center;">5.216</td>
623
+ </tr>
624
+ <tr>
625
+ <td style="text-align: left;">Russian</td>
626
+ <td style="text-align: center;">5.957</td>
627
+ <td style="text-align: center;">4.535</td>
628
+ <td style="text-align: center;">4.458</td>
629
+ <td style="text-align: center;"><strong>3.212</strong></td>
630
+ <td style="text-align: center;">4.281</td>
631
+ <td style="text-align: center;">3.878</td>
632
+ </tr>
633
+ <tr style="border-top: 1px solid #ddd;">
634
+ <td colspan="7" style="text-align: center; border-bottom: 1px solid #ddd;"><em>Speaker Similarity</em></td>
635
+ </tr>
636
+ <tr>
637
+ <td style="text-align: left;">Chinese</td>
638
+ <td style="text-align: center;">0.797</td>
639
+ <td style="text-align: center;">0.796</td>
640
+ <td style="text-align: center;"><strong>0.811</strong></td>
641
+ <td style="text-align: center;">0.799</td>
642
+ <td style="text-align: center;">0.780</td>
643
+ <td style="text-align: center;">0.677</td>
644
+ </tr>
645
+ <tr>
646
+ <td style="text-align: left;">English</td>
647
+ <td style="text-align: center;">0.811</td>
648
+ <td style="text-align: center;">0.815</td>
649
+ <td style="text-align: center;"><strong>0.829</strong></td>
650
+ <td style="text-align: center;">0.775</td>
651
+ <td style="text-align: center;">0.756</td>
652
+ <td style="text-align: center;">0.613</td>
653
+ </tr>
654
+ <tr>
655
+ <td style="text-align: left;">German</td>
656
+ <td style="text-align: center;">0.749</td>
657
+ <td style="text-align: center;">0.737</td>
658
+ <td style="text-align: center;">0.769</td>
659
+ <td style="text-align: center;"><strong>0.775</strong></td>
660
+ <td style="text-align: center;">0.733</td>
661
+ <td style="text-align: center;">0.614</td>
662
+ </tr>
663
+ <tr>
664
+ <td style="text-align: left;">Italian</td>
665
+ <td style="text-align: center;">0.722</td>
666
+ <td style="text-align: center;">0.718</td>
667
+ <td style="text-align: center;">0.792</td>
668
+ <td style="text-align: center;"><strong>0.817</strong></td>
669
+ <td style="text-align: center;">0.699</td>
670
+ <td style="text-align: center;">0.579</td>
671
+ </tr>
672
+ <tr>
673
+ <td style="text-align: left;">Portuguese</td>
674
+ <td style="text-align: center;">0.790</td>
675
+ <td style="text-align: center;">0.783</td>
676
+ <td style="text-align: center;">0.794</td>
677
+ <td style="text-align: center;"><strong>0.817</strong></td>
678
+ <td style="text-align: center;">0.805</td>
679
+ <td style="text-align: center;">0.711</td>
680
+ </tr>
681
+ <tr>
682
+ <td style="text-align: left;">Spanish</td>
683
+ <td style="text-align: center;">0.732</td>
684
+ <td style="text-align: center;">0.731</td>
685
+ <td style="text-align: center;">0.812</td>
686
+ <td style="text-align: center;"><strong>0.814</strong></td>
687
+ <td style="text-align: center;">0.762</td>
688
+ <td style="text-align: center;">0.615</td>
689
+ </tr>
690
+ <tr>
691
+ <td style="text-align: left;">Japanese</td>
692
+ <td style="text-align: center;"><strong>0.810</strong></td>
693
+ <td style="text-align: center;">0.807</td>
694
+ <td style="text-align: center;">0.798</td>
695
+ <td style="text-align: center;">0.788</td>
696
+ <td style="text-align: center;">0.776</td>
697
+ <td style="text-align: center;">0.738</td>
698
+ </tr>
699
+ <tr>
700
+ <td style="text-align: left;">Korean</td>
701
+ <td style="text-align: center;"><strong>0.824</strong></td>
702
+ <td style="text-align: center;">0.814</td>
703
+ <td style="text-align: center;">0.812</td>
704
+ <td style="text-align: center;">0.799</td>
705
+ <td style="text-align: center;">0.779</td>
706
+ <td style="text-align: center;">0.700</td>
707
+ </tr>
708
+ <tr>
709
+ <td style="text-align: left;">French</td>
710
+ <td style="text-align: center;">0.698</td>
711
+ <td style="text-align: center;">0.703</td>
712
+ <td style="text-align: center;">0.700</td>
713
+ <td style="text-align: center;"><strong>0.714</strong></td>
714
+ <td style="text-align: center;">0.628</td>
715
+ <td style="text-align: center;">0.535</td>
716
+ </tr>
717
+ <tr>
718
+ <td style="text-align: left;">Russian</td>
719
+ <td style="text-align: center;">0.734</td>
720
+ <td style="text-align: center;">0.744</td>
721
+ <td style="text-align: center;">0.781</td>
722
+ <td style="text-align: center;"><strong>0.792</strong></td>
723
+ <td style="text-align: center;">0.761</td>
724
+ <td style="text-align: center;">0.676</td>
725
+ </tr>
726
+ </tbody>
727
+ </table>
728
+
729
+ <br>
730
+
731
+ *Cross-lingual speech generation on the Cross-Lingual benchmark. Performance is measured by Mixed Error Rate (WER for English, CER for others, ↓).*
732
+
733
+ <table>
734
+ <thead>
735
+ <tr>
736
+ <th style="text-align: left;">Task</th>
737
+ <th style="text-align: center;">Qwen3-TTS-25Hz-1.7B-Base</th>
738
+ <th style="text-align: center;">Qwen3-TTS-12Hz-1.7B-Base</th>
739
+ <th style="text-align: center;">CosyVoice3</th>
740
+ <th style="text-align: center;">CosyVoice2</th>
741
+ </tr>
742
+ </thead>
743
+ <tbody>
744
+ <tr>
745
+ <td style="text-align: left;">en-to-zh</td>
746
+ <td style="text-align: center;">5.66</td>
747
+ <td style="text-align: center;"><strong>4.77</strong></td>
748
+ <td style="text-align: center;">5.09</td>
749
+ <td style="text-align: center;">13.5</td>
750
+ </tr>
751
+ <tr>
752
+ <td style="text-align: left;">ja-to-zh</td>
753
+ <td style="text-align: center;">3.92</td>
754
+ <td style="text-align: center;">3.43</td>
755
+ <td style="text-align: center;"><strong>3.05</strong></td>
756
+ <td style="text-align: center;">48.1</td>
757
+ </tr>
758
+ <tr>
759
+ <td style="text-align: left;">ko-to-zh</td>
760
+ <td style="text-align: center;">1.14</td>
761
+ <td style="text-align: center;">1.08</td>
762
+ <td style="text-align: center;"><strong>1.06</strong></td>
763
+ <td style="text-align: center;">7.70</td>
764
+ </tr>
765
+ <tr style="border-top: 1px solid #ddd;">
766
+ <td style="text-align: left;">zh-to-en</td>
767
+ <td style="text-align: center;">2.91</td>
768
+ <td style="text-align: center;"><strong>2.77</strong></td>
769
+ <td style="text-align: center;">2.98</td>
770
+ <td style="text-align: center;">6.47</td>
771
+ </tr>
772
+ <tr>
773
+ <td style="text-align: left;">ja-to-en</td>
774
+ <td style="text-align: center;">3.95</td>
775
+ <td style="text-align: center;"><strong>3.04</strong></td>
776
+ <td style="text-align: center;">4.20</td>
777
+ <td style="text-align: center;">17.1</td>
778
+ </tr>
779
+ <tr>
780
+ <td style="text-align: left;">ko-to-en</td>
781
+ <td style="text-align: center;">3.48</td>
782
+ <td style="text-align: center;"><strong>3.09</strong></td>
783
+ <td style="text-align: center;">4.19</td>
784
+ <td style="text-align: center;">11.2</td>
785
+ </tr>
786
+ <tr style="border-top: 1px solid #ddd;">
787
+ <td style="text-align: left;">zh-to-ja</td>
788
+ <td style="text-align: center;">9.29</td>
789
+ <td style="text-align: center;">8.40</td>
790
+ <td style="text-align: center;"><strong>7.08</strong></td>
791
+ <td style="text-align: center;">13.1</td>
792
+ </tr>
793
+ <tr>
794
+ <td style="text-align: left;">en-to-ja</td>
795
+ <td style="text-align: center;">7.74</td>
796
+ <td style="text-align: center;">7.21</td>
797
+ <td style="text-align: center;"><strong>6.80</strong></td>
798
+ <td style="text-align: center;">14.9</td>
799
+ </tr>
800
+ <tr>
801
+ <td style="text-align: left;">ko-to-ja</td>
802
+ <td style="text-align: center;">4.17</td>
803
+ <td style="text-align: center;"><strong>3.67</strong></td>
804
+ <td style="text-align: center;">3.93</td>
805
+ <td style="text-align: center;">5.86</td>
806
+ </tr>
807
+ <tr style="border-top: 1px solid #ddd;">
808
+ <td style="text-align: left;">zh-to-ko</td>
809
+ <td style="text-align: center;">8.12</td>
810
+ <td style="text-align: center;"><strong>4.82</strong></td>
811
+ <td style="text-align: center;">14.4</td>
812
+ <td style="text-align: center;">24.8</td>
813
+ </tr>
814
+ <tr>
815
+ <td style="text-align: left;">en-to-ko</td>
816
+ <td style="text-align: center;">6.83</td>
817
+ <td style="text-align: center;"><strong>5.14</strong></td>
818
+ <td style="text-align: center;">5.87</td>
819
+ <td style="text-align: center;">21.9</td>
820
+ </tr>
821
+ <tr>
822
+ <td style="text-align: left;">ja-to-ko</td>
823
+ <td style="text-align: center;">6.86</td>
824
+ <td style="text-align: center;"><strong>5.59</strong></td>
825
+ <td style="text-align: center;">7.92</td>
826
+ <td style="text-align: center;">21.5</td>
827
+ </tr>
828
+ </tbody>
829
+ </table>
830
+
831
+ <br>
832
+
833
+ *Controllable speech generation on InstructTTSEval. Performance is measured by Attribute Perception and Synthesis accuracy (APS), Description-Speech Consistency (DSD), and Response Precision (RP).*
834
+
835
+ <table>
836
+ <thead>
837
+ <tr>
838
+ <th rowspan="2" style="text-align: left; vertical-align: bottom;">Type</th>
839
+ <th rowspan="2" style="text-align: left; vertical-align: bottom;">Model</th>
840
+ <th colspan="3" style="text-align: center;">InstructTTSEval-ZH</th>
841
+ <th colspan="3" style="text-align: center;">InstructTTSEval-EN</th>
842
+ </tr>
843
+ <tr>
844
+ <th style="text-align: center;">APS (↑)</th>
845
+ <th style="text-align: center;">DSD (↑)</th>
846
+ <th style="text-align: center;">RP (↑)</th>
847
+ <th style="text-align: center;">APS (↑)</th>
848
+ <th style="text-align: center;">DSD (↑)</th>
849
+ <th style="text-align: center;">RP (↑)</th>
850
+ </tr>
851
+ </thead>
852
+ <tbody>
853
+ <tr>
854
+ <td rowspan="5" style="text-align: left; vertical-align: middle;"><em>Target<br>Speaker</em></td>
855
+ <td style="text-align: left;">Gemini-flash</td>
856
+ <td style="text-align: center;">88.2</td>
857
+ <td style="text-align: center;"><strong>90.9</strong></td>
858
+ <td style="text-align: center;"><strong>77.3</strong></td>
859
+ <td style="text-align: center;"><strong>92.3</strong></td>
860
+ <td style="text-align: center;"><strong>93.8</strong></td>
861
+ <td style="text-align: center;"><strong>80.1</strong></td>
862
+ </tr>
863
+ <tr>
864
+ <td style="text-align: left;">Gemini-pro</td>
865
+ <td style="text-align: center;"><strong>89.0</strong></td>
866
+ <td style="text-align: center;">90.1</td>
867
+ <td style="text-align: center;">75.5</td>
868
+ <td style="text-align: center;">87.6</td>
869
+ <td style="text-align: center;">86.0</td>
870
+ <td style="text-align: center;">67.2</td>
871
+ </tr>
872
+ <tr>
873
+ <td style="text-align: left;">Qwen3TTS-25Hz-1.7B-CustomVoice</td>
874
+ <td style="text-align: center;">83.1</td>
875
+ <td style="text-align: center;">75.0</td>
876
+ <td style="text-align: center;">63.0</td>
877
+ <td style="text-align: center;">79.0</td>
878
+ <td style="text-align: center;">82.8</td>
879
+ <td style="text-align: center;">69.3</td>
880
+ </tr>
881
+ <tr>
882
+ <td style="text-align: left;">Qwen3TTS-12Hz-1.7B-CustomVoice</td>
883
+ <td style="text-align: center;">83.0</td>
884
+ <td style="text-align: center;">77.8</td>
885
+ <td style="text-align: center;">61.2</td>
886
+ <td style="text-align: center;">77.3</td>
887
+ <td style="text-align: center;">77.1</td>
888
+ <td style="text-align: center;">63.7</td>
889
+ </tr>
890
+ <tr>
891
+ <td style="text-align: left;">GPT-4o-mini-tts</td>
892
+ <td style="text-align: center;">54.9</td>
893
+ <td style="text-align: center;">52.3</td>
894
+ <td style="text-align: center;">46.0</td>
895
+ <td style="text-align: center;">76.4</td>
896
+ <td style="text-align: center;">74.3</td>
897
+ <td style="text-align: center;">54.8</td>
898
+ </tr>
899
+ <tr style="border-top: 1px solid #ddd;">
900
+ <td rowspan="9" style="text-align: left; vertical-align: middle;"><em>Voice<br>Design</em></td>
901
+ <td style="text-align: left;">Qwen3TTS-12Hz-1.7B-VD</td>
902
+ <td style="text-align: center;"><strong>85.2</strong></td>
903
+ <td style="text-align: center;"><strong>81.1</strong></td>
904
+ <td style="text-align: center;"><strong>65.1</strong></td>
905
+ <td style="text-align: center;">82.9</td>
906
+ <td style="text-align: center;"><strong>82.4</strong></td>
907
+ <td style="text-align: center;"><strong>68.4</strong></td>
908
+ </tr>
909
+ <tr>
910
+ <td style="text-align: left;">Mimo-Audio-7B-Instruct (Zhang et al., 2025b)</td>
911
+ <td style="text-align: center;">75.7</td>
912
+ <td style="text-align: center;">74.3</td>
913
+ <td style="text-align: center;">61.5</td>
914
+ <td style="text-align: center;">80.6</td>
915
+ <td style="text-align: center;">77.6</td>
916
+ <td style="text-align: center;">59.5</td>
917
+ </tr>
918
+ <tr>
919
+ <td style="text-align: left;">VoiceSculptor (Hu et al., 2026)</td>
920
+ <td style="text-align: center;">75.7</td>
921
+ <td style="text-align: center;">64.7</td>
922
+ <td style="text-align: center;">61.5</td>
923
+ <td style="text-align: center;">-</td>
924
+ <td style="text-align: center;">-</td>
925
+ <td style="text-align: center;">-</td>
926
+ </tr>
927
+ <tr>
928
+ <td style="text-align: left;">Hume</td>
929
+ <td style="text-align: center;">-</td>
930
+ <td style="text-align: center;">-</td>
931
+ <td style="text-align: center;">-</td>
932
+ <td style="text-align: center;"><strong>83.0</strong></td>
933
+ <td style="text-align: center;">75.3</td>
934
+ <td style="text-align: center;">54.3</td>
935
+ </tr>
936
+ <tr>
937
+ <td style="text-align: left;">VoxInstruct (Zhou et al., 2024)</td>
938
+ <td style="text-align: center;">47.5</td>
939
+ <td style="text-align: center;">52.3</td>
940
+ <td style="text-align: center;">42.6</td>
941
+ <td style="text-align: center;">54.9</td>
942
+ <td style="text-align: center;">57.0</td>
943
+ <td style="text-align: center;">39.3</td>
944
+ </tr>
945
+ <tr>
946
+ <td style="text-align: left;">Parler-tts-mini (Lyth & King, 2024)</td>
947
+ <td style="text-align: center;">-</td>
948
+ <td style="text-align: center;">-</td>
949
+ <td style="text-align: center;">-</td>
950
+ <td style="text-align: center;">63.4</td>
951
+ <td style="text-align: center;">48.7</td>
952
+ <td style="text-align: center;">28.6</td>
953
+ </tr>
954
+ <tr>
955
+ <td style="text-align: left;">Parler-tts-large (Lyth & King, 2024)</td>
956
+ <td style="text-align: center;">-</td>
957
+ <td style="text-align: center;">-</td>
958
+ <td style="text-align: center;">-</td>
959
+ <td style="text-align: center;">60.0</td>
960
+ <td style="text-align: center;">45.9</td>
961
+ <td style="text-align: center;">31.2</td>
962
+ </tr>
963
+ <tr>
964
+ <td style="text-align: left;">PromptTTS (Guo et al., 2023)</td>
965
+ <td style="text-align: center;">-</td>
966
+ <td style="text-align: center;">-</td>
967
+ <td style="text-align: center;">-</td>
968
+ <td style="text-align: center;">64.3</td>
969
+ <td style="text-align: center;">47.2</td>
970
+ <td style="text-align: center;">31.4</td>
971
+ </tr>
972
+ <tr>
973
+ <td style="text-align: left;">PromptStyle (Liu et al., 2023)</td>
974
+ <td style="text-align: center;">-</td>
975
+ <td style="text-align: center;">-</td>
976
+ <td style="text-align: center;">-</td>
977
+ <td style="text-align: center;">57.4</td>
978
+ <td style="text-align: center;">46.4</td>
979
+ <td style="text-align: center;">30.9</td>
980
+ </tr>
981
+ </tbody>
982
+ </table>
983
+
984
+ <br>
985
+
986
+ *Target-Speaker Multilingual Speech Generation on the TTS multilingual test set. Performance is measured by Word Error Rate (WER, ↓).*
987
+
988
+ <table>
989
+ <thead>
990
+ <tr>
991
+ <th rowspan="2" style="text-align: left; vertical-align: bottom;">Language</th>
992
+ <th colspan="2" style="text-align: center;">Qwen3-TTS-25Hz</th>
993
+ <th colspan="2" style="text-align: center;">Qwen3-TTS-12Hz</th>
994
+ <th rowspan="2" style="text-align: center; vertical-align: bottom;">GPT-4o-Audio<br>Preview</th>
995
+ </tr>
996
+ <tr>
997
+ <th style="text-align: center;">0.6B-CustomVoice</th>
998
+ <th style="text-align: center;">1.7B-CustomVoice</th>
999
+ <th style="text-align: center;">0.6B-CustomVoice</th>
1000
+ <th style="text-align: center;">1.7B-CustomVoice</th>
1001
+ </tr>
1002
+ </thead>
1003
+ <tbody>
1004
+ <tr>
1005
+ <td style="text-align: left;">Chinese</td>
1006
+ <td style="text-align: center;">0.874</td>
1007
+ <td style="text-align: center;"><strong>0.708</strong></td>
1008
+ <td style="text-align: center;">0.944</td>
1009
+ <td style="text-align: center;">0.903</td>
1010
+ <td style="text-align: center;">3.519</td>
1011
+ </tr>
1012
+ <tr>
1013
+ <td style="text-align: left;">English</td>
1014
+ <td style="text-align: center;">1.332</td>
1015
+ <td style="text-align: center;">0.936</td>
1016
+ <td style="text-align: center;">1.188</td>
1017
+ <td style="text-align: center;"><strong>0.899</strong></td>
1018
+ <td style="text-align: center;">2.197</td>
1019
+ </tr>
1020
+ <tr>
1021
+ <td style="text-align: left;">German</td>
1022
+ <td style="text-align: center;">0.990</td>
1023
+ <td style="text-align: center;"><strong>0.634</strong></td>
1024
+ <td style="text-align: center;">2.722</td>
1025
+ <td style="text-align: center;">1.057</td>
1026
+ <td style="text-align: center;">1.161</td>
1027
+ </tr>
1028
+ <tr>
1029
+ <td style="text-align: left;">Italian</td>
1030
+ <td style="text-align: center;">1.861</td>
1031
+ <td style="text-align: center;">1.271</td>
1032
+ <td style="text-align: center;">2.545</td>
1033
+ <td style="text-align: center;">1.362</td>
1034
+ <td style="text-align: center;"><strong>1.194</strong></td>
1035
+ </tr>
1036
+ <tr>
1037
+ <td style="text-align: left;">Portuguese</td>
1038
+ <td style="text-align: center;">1.728</td>
1039
+ <td style="text-align: center;">1.854</td>
1040
+ <td style="text-align: center;">3.219</td>
1041
+ <td style="text-align: center;">2.681</td>
1042
+ <td style="text-align: center;"><strong>1.504</strong></td>
1043
+ </tr>
1044
+ <tr>
1045
+ <td style="text-align: left;">Spanish</td>
1046
+ <td style="text-align: center;">1.309</td>
1047
+ <td style="text-align: center;">1.284</td>
1048
+ <td style="text-align: center;"><strong>1.154</strong></td>
1049
+ <td style="text-align: center;">1.330</td>
1050
+ <td style="text-align: center;">4.000</td>
1051
+ </tr>
1052
+ <tr>
1053
+ <td style="text-align: left;">Japanese</td>
1054
+ <td style="text-align: center;"><strong>3.875</strong></td>
1055
+ <td style="text-align: center;">4.518</td>
1056
+ <td style="text-align: center;">6.877</td>
1057
+ <td style="text-align: center;">4.924</td>
1058
+ <td style="text-align: center;">5.001</td>
1059
+ </tr>
1060
+ <tr>
1061
+ <td style="text-align: left;">Korean</td>
1062
+ <td style="text-align: center;">2.202</td>
1063
+ <td style="text-align: center;">2.274</td>
1064
+ <td style="text-align: center;">3.053</td>
1065
+ <td style="text-align: center;"><strong>1.741</strong></td>
1066
+ <td style="text-align: center;">2.763</td>
1067
+ </tr>
1068
+ <tr>
1069
+ <td style="text-align: left;">French</td>
1070
+ <td style="text-align: center;">3.865</td>
1071
+ <td style="text-align: center;"><strong>3.080</strong></td>
1072
+ <td style="text-align: center;">3.841</td>
1073
+ <td style="text-align: center;">3.781</td>
1074
+ <td style="text-align: center;">3.605</td>
1075
+ </tr>
1076
+ <tr>
1077
+ <td style="text-align: left;">Russian</td>
1078
+ <td style="text-align: center;">6.529</td>
1079
+ <td style="text-align: center;"><strong>4.444</strong></td>
1080
+ <td style="text-align: center;">5.809</td>
1081
+ <td style="text-align: center;">4.734</td>
1082
+ <td style="text-align: center;">5.250</td>
1083
+ </tr>
1084
+ </tbody>
1085
+ </table>
1086
+
1087
+ <br>
1088
+
1089
+ *Long speech generation results. Performance is measured by Word Error Rate (WER, ↓).*
1090
+
1091
+ <table>
1092
+ <thead>
1093
+ <tr>
1094
+ <th style="text-align: center;">Datasets</th>
1095
+ <th style="text-align: left;">Model</th>
1096
+ <th colspan="2" style="text-align: center;">Performance</th>
1097
+ </tr>
1098
+ <tr style="border-bottom: 1px solid #ddd; border-top: 1px solid #ddd;">
1099
+ <td colspan="4" style="text-align: center;"><em>Content Consistency</em></td>
1100
+ </tr>
1101
+ </thead>
1102
+ <tbody>
1103
+ <tr>
1104
+ <td rowspan="5" style="text-align: center; vertical-align: middle;"><em>long-zh</em> | <em>long-en</em></td>
1105
+ <td style="text-align: left;">Higgs-Audio-v2 (chunk) (Boson AI, 2025)</td>
1106
+ <td style="text-align: center;">5.505</td>
1107
+ <td style="text-align: center;">6.917</td>
1108
+ </tr>
1109
+ <tr>
1110
+ <td style="text-align: left;">VibeVoice (Peng et al., 2025)</td>
1111
+ <td style="text-align: center;">22.619</td>
1112
+ <td style="text-align: center;">1.780</td>
1113
+ </tr>
1114
+ <tr>
1115
+ <td style="text-align: left;">VoxCPM (Zhou et al., 2025)</td>
1116
+ <td style="text-align: center;">4.835</td>
1117
+ <td style="text-align: center;">7.474</td>
1118
+ </tr>
1119
+ <tr>
1120
+ <td style="text-align: left;">Qwen3-TTS-25Hz-1.7B-CustomVoice</td>
1121
+ <td style="text-align: center;"><strong>1.517</strong></td>
1122
+ <td style="text-align: center;"><strong>1.225</strong></td>
1123
+ </tr>
1124
+ <tr>
1125
+ <td style="text-align: left;">Qwen3-TTS-12Hz-1.7B-CustomVoice</td>
1126
+ <td style="text-align: center;">2.356</td>
1127
+ <td style="text-align: center;">2.812</td>
1128
+ </tr>
1129
+ </tbody>
1130
+ </table>
1131
+ </details>
1132
+
1133
+
1134
+ <details>
1135
+ <summary>Speech Tokenizer Benchmarks</summary>
1136
+
1137
+ *Comparison between different supervised semantic speech tokenizers on ASR Task.*
1138
+
1139
+ <table>
1140
+ <thead>
1141
+ <tr>
1142
+ <th style="text-align: left;">Model</th>
1143
+ <th style="text-align: center;">Codebook Size</th>
1144
+ <th style="text-align: center;">FPS</th>
1145
+ <th style="text-align: center;">C.V. EN</th>
1146
+ <th style="text-align: center;">C.V. CN</th>
1147
+ <th style="text-align: center;">Fluers EN</th>
1148
+ <th style="text-align: center;">Fluers CN</th>
1149
+ </tr>
1150
+ </thead>
1151
+ <tbody>
1152
+ <tr>
1153
+ <td style="text-align: left;">S3 Tokenizer(VQ) (Du et al., 2024a)</td>
1154
+ <td style="text-align: center;">4096</td>
1155
+ <td style="text-align: center;">50</td>
1156
+ <td style="text-align: center;">12.06</td>
1157
+ <td style="text-align: center;">15.38</td>
1158
+ <td style="text-align: center;">-</td>
1159
+ <td style="text-align: center;">-</td>
1160
+ </tr>
1161
+ <tr>
1162
+ <td style="text-align: left;">S3 Tokenizer(VQ) (Du et al., 2024a)</td>
1163
+ <td style="text-align: center;">4096</td>
1164
+ <td style="text-align: center;">25</td>
1165
+ <td style="text-align: center;">11.56</td>
1166
+ <td style="text-align: center;">18.26</td>
1167
+ <td style="text-align: center;">7.65</td>
1168
+ <td style="text-align: center;">5.03</td>
1169
+ </tr>
1170
+ <tr>
1171
+ <td style="text-align: left;">S3 Tokenizer(FSQ) (Du et al., 2024a)</td>
1172
+ <td style="text-align: center;">6561</td>
1173
+ <td style="text-align: center;">25</td>
1174
+ <td style="text-align: center;">10.67</td>
1175
+ <td style="text-align: center;"><strong>7.29</strong></td>
1176
+ <td style="text-align: center;">6.58</td>
1177
+ <td style="text-align: center;">4.43</td>
1178
+ </tr>
1179
+ <tr>
1180
+ <td style="text-align: left;">Qwen-TTS-Tokenizer-25Hz (Stage 1)</td>
1181
+ <td style="text-align: center;">32768</td>
1182
+ <td style="text-align: center;">25</td>
1183
+ <td style="text-align: center;"><strong>7.51</strong></td>
1184
+ <td style="text-align: center;">10.73</td>
1185
+ <td style="text-align: center;"><strong>3.07</strong></td>
1186
+ <td style="text-align: center;"><strong>4.23</strong></td>
1187
+ </tr>
1188
+ <tr>
1189
+ <td style="text-align: left;">Qwen-TTS-Tokenizer-25Hz (Stage 2)</td>
1190
+ <td style="text-align: center;">32768</td>
1191
+ <td style="text-align: center;">25</td>
1192
+ <td style="text-align: center;">10.40</td>
1193
+ <td style="text-align: center;">14.99</td>
1194
+ <td style="text-align: center;">4.14</td>
1195
+ <td style="text-align: center;">4.67</td>
1196
+ </tr>
1197
+ </tbody>
1198
+ </table>
1199
+
1200
+ <br>
1201
+
1202
+ *Comparison between different semantic-related speech tokenizers.*
1203
+
1204
+ <table>
1205
+ <thead>
1206
+ <tr>
1207
+ <th style="text-align: left;">Model</th>
1208
+ <th style="text-align: center;">NQ</th>
1209
+ <th style="text-align: center;">Codebook Size</th>
1210
+ <th style="text-align: center;">FPS</th>
1211
+ <th style="text-align: center;">PESQ_WB</th>
1212
+ <th style="text-align: center;">PESQ_NB</th>
1213
+ <th style="text-align: center;">STOI</th>
1214
+ <th style="text-align: center;">UTMOS</th>
1215
+ <th style="text-align: center;">SIM</th>
1216
+ </tr>
1217
+ </thead>
1218
+ <tbody>
1219
+ <tr>
1220
+ <td style="text-align: left;">SpeechTokenizer (Zhang et al., 2023a)</td>
1221
+ <td style="text-align: center;">8</td>
1222
+ <td style="text-align: center;">1024</td>
1223
+ <td style="text-align: center;">50</td>
1224
+ <td style="text-align: center;">2.60</td>
1225
+ <td style="text-align: center;">3.05</td>
1226
+ <td style="text-align: center;">0.92</td>
1227
+ <td style="text-align: center;">3.90</td>
1228
+ <td style="text-align: center;">0.85</td>
1229
+ </tr>
1230
+ <tr>
1231
+ <td style="text-align: left;">X-codec (Ye et al., 2025a)</td>
1232
+ <td style="text-align: center;">2</td>
1233
+ <td style="text-align: center;">1024</td>
1234
+ <td style="text-align: center;">50</td>
1235
+ <td style="text-align: center;">2.68</td>
1236
+ <td style="text-align: center;">3.27</td>
1237
+ <td style="text-align: center;">0.86</td>
1238
+ <td style="text-align: center;">4.11</td>
1239
+ <td style="text-align: center;">0.84</td>
1240
+ </tr>
1241
+ <tr>
1242
+ <td style="text-align: left;">X-codec 2 (Ye et al., 2025b)</td>
1243
+ <td style="text-align: center;">1</td>
1244
+ <td style="text-align: center;">65536</td>
1245
+ <td style="text-align: center;">50</td>
1246
+ <td style="text-align: center;">2.43</td>
1247
+ <td style="text-align: center;">3.04</td>
1248
+ <td style="text-align: center;">0.92</td>
1249
+ <td style="text-align: center;">4.13</td>
1250
+ <td style="text-align: center;">0.82</td>
1251
+ </tr>
1252
+ <tr>
1253
+ <td style="text-align: left;">XY-Tokenizer (Gong et al., 2025)</td>
1254
+ <td style="text-align: center;">8</td>
1255
+ <td style="text-align: center;">1024</td>
1256
+ <td style="text-align: center;">12.5</td>
1257
+ <td style="text-align: center;">2.41</td>
1258
+ <td style="text-align: center;">3.00</td>
1259
+ <td style="text-align: center;">0.91</td>
1260
+ <td style="text-align: center;">3.98</td>
1261
+ <td style="text-align: center;">0.83</td>
1262
+ </tr>
1263
+ <tr>
1264
+ <td style="text-align: left;">Mimi (Défossez et al., 2024)</td>
1265
+ <td style="text-align: center;">16</td>
1266
+ <td style="text-align: center;">2048</td>
1267
+ <td style="text-align: center;">12.5</td>
1268
+ <td style="text-align: center;">2.88</td>
1269
+ <td style="text-align: center;">3.42</td>
1270
+ <td style="text-align: center;">0.94</td>
1271
+ <td style="text-align: center;">3.87</td>
1272
+ <td style="text-align: center;">0.87</td>
1273
+ </tr>
1274
+ <tr>
1275
+ <td style="text-align: left;">FireredTTS 2 Tokenizer (Xie et al., 2025)</td>
1276
+ <td style="text-align: center;">16</td>
1277
+ <td style="text-align: center;">2048</td>
1278
+ <td style="text-align: center;">12.5</td>
1279
+ <td style="text-align: center;">2.73</td>
1280
+ <td style="text-align: center;">3.28</td>
1281
+ <td style="text-align: center;">0.94</td>
1282
+ <td style="text-align: center;">3.88</td>
1283
+ <td style="text-align: center;">0.87</td>
1284
+ </tr>
1285
+ <tr>
1286
+ <td style="text-align: left;">Qwen-TTS-Tokenizer-12Hz</td>
1287
+ <td style="text-align: center;">16</td>
1288
+ <td style="text-align: center;">2048</td>
1289
+ <td style="text-align: center;">12.5</td>
1290
+ <td style="text-align: center;"><strong>3.21</strong></td>
1291
+ <td style="text-align: center;"><strong>3.68</strong></td>
1292
+ <td style="text-align: center;"><strong>0.96</strong></td>
1293
+ <td style="text-align: center;"><strong>4.16</strong></td>
1294
+ <td style="text-align: center;"><strong>0.95</strong></td>
1295
+ </tr>
1296
+ </tbody>
1297
+ </table>
1298
+
1299
+ </details>
1300
+
1301
+
1302
+ ## Citation
1303
+
1304
+ If you find our paper and code useful in your research, please consider giving a star :star: and citation :pencil: :)
1305
+
1306
+ ```BibTeX
1307
+ @article{Qwen3-TTS,
1308
+ title={Qwen3-TTS Technical Report},
1309
+ author={Hangrui Hu and Xinfa Zhu and Ting He and Dake Guo and Bin Zhang and Xiong Wang and Zhifang Guo and Ziyue Jiang and Hongkun Hao and Zishan Guo and Xinyu Zhang and Pei Zhang and Baosong Yang and Jin Xu and Jingren Zhou and Junyang Lin},
1310
+ journal={arXiv preprint arXiv:2601.15621},
1311
+ year={2026}
1312
+ }
1313
+ ```
1314
+
1315
+ <br>
qwen3-tts/config.json ADDED
@@ -0,0 +1,167 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3TTSForConditionalGeneration"
4
+ ],
5
+ "assistant_token_id": 77091,
6
+ "im_end_token_id": 151645,
7
+ "im_start_token_id": 151644,
8
+ "tts_bos_token_id": 151672,
9
+ "tts_eos_token_id": 151673,
10
+ "tts_pad_token_id": 151671,
11
+ "model_type": "qwen3_tts",
12
+ "tokenizer_type": "qwen3_tts_tokenizer_12hz",
13
+ "tts_model_size": "1b7",
14
+ "tts_model_type": "base",
15
+ "speaker_encoder_config": {
16
+ "enc_dim": 2048,
17
+ "sample_rate": 24000
18
+ },
19
+ "talker_config": {
20
+ "attention_bias": false,
21
+ "attention_dropout": 0,
22
+ "code_predictor_config": {
23
+ "_name_or_path": "",
24
+ "add_cross_attention": false,
25
+ "architectures": null,
26
+ "attention_bias": false,
27
+ "attention_dropout": 0,
28
+ "bad_words_ids": null,
29
+ "begin_suppress_tokens": null,
30
+ "bos_token_id": null,
31
+ "chunk_size_feed_forward": 0,
32
+ "cross_attention_hidden_size": null,
33
+ "decoder_start_token_id": null,
34
+ "diversity_penalty": 0.0,
35
+ "do_sample": false,
36
+ "early_stopping": false,
37
+ "encoder_no_repeat_ngram_size": 0,
38
+ "eos_token_id": null,
39
+ "exponential_decay_length_penalty": null,
40
+ "finetuning_task": null,
41
+ "forced_bos_token_id": null,
42
+ "forced_eos_token_id": null,
43
+ "head_dim": 128,
44
+ "hidden_act": "silu",
45
+ "hidden_size": 1024,
46
+ "id2label": {
47
+ "0": "LABEL_0",
48
+ "1": "LABEL_1"
49
+ },
50
+ "initializer_range": 0.02,
51
+ "intermediate_size": 3072,
52
+ "is_decoder": false,
53
+ "is_encoder_decoder": false,
54
+ "label2id": {
55
+ "LABEL_0": 0,
56
+ "LABEL_1": 1
57
+ },
58
+ "layer_types": [
59
+ "full_attention",
60
+ "full_attention",
61
+ "full_attention",
62
+ "full_attention",
63
+ "full_attention"
64
+ ],
65
+ "length_penalty": 1.0,
66
+ "max_length": 20,
67
+ "max_position_embeddings": 65536,
68
+ "max_window_layers": 28,
69
+ "min_length": 0,
70
+ "model_type": "qwen3_tts_talker_code_predictor",
71
+ "no_repeat_ngram_size": 0,
72
+ "num_attention_heads": 16,
73
+ "num_beam_groups": 1,
74
+ "num_beams": 1,
75
+ "num_code_groups": 16,
76
+ "num_hidden_layers": 5,
77
+ "num_key_value_heads": 8,
78
+ "num_return_sequences": 1,
79
+ "output_attentions": false,
80
+ "output_hidden_states": false,
81
+ "output_scores": false,
82
+ "pad_token_id": null,
83
+ "prefix": null,
84
+ "problem_type": null,
85
+ "pruned_heads": {},
86
+ "remove_invalid_values": false,
87
+ "repetition_penalty": 1.0,
88
+ "return_dict": true,
89
+ "return_dict_in_generate": false,
90
+ "rms_norm_eps": 1e-06,
91
+ "rope_scaling": null,
92
+ "rope_theta": 1000000,
93
+ "sep_token_id": null,
94
+ "sliding_window": null,
95
+ "suppress_tokens": null,
96
+ "task_specific_params": null,
97
+ "temperature": 1.0,
98
+ "tf_legacy_loss": false,
99
+ "tie_encoder_decoder": false,
100
+ "tie_word_embeddings": false,
101
+ "tokenizer_class": null,
102
+ "top_k": 50,
103
+ "top_p": 1.0,
104
+ "dtype": null,
105
+ "torchscript": false,
106
+ "typical_p": 1.0,
107
+ "use_bfloat16": false,
108
+ "use_cache": true,
109
+ "use_sliding_window": false,
110
+ "vocab_size": 2048
111
+ },
112
+ "codec_bos_id": 2149,
113
+ "codec_eos_token_id": 2150,
114
+ "codec_think_id": 2154,
115
+ "codec_language_id": {
116
+ "chinese": 2055,
117
+ "english": 2050,
118
+ "german": 2053,
119
+ "italian": 2070,
120
+ "portuguese": 2071,
121
+ "spanish": 2054,
122
+ "japanese": 2058,
123
+ "korean": 2064,
124
+ "french": 2061,
125
+ "russian": 2069
126
+ },
127
+ "codec_nothink_id": 2155,
128
+ "codec_pad_id": 2148,
129
+ "codec_think_bos_id": 2156,
130
+ "codec_think_eos_id": 2157,
131
+ "spk_id": {
132
+ },
133
+ "spk_is_dialect": {
134
+ },
135
+ "head_dim": 128,
136
+ "hidden_act": "silu",
137
+ "hidden_size": 2048,
138
+ "initializer_range": 0.02,
139
+ "intermediate_size": 6144,
140
+ "max_position_embeddings": 32768,
141
+ "model_type": "qwen3_tts_talker",
142
+ "num_attention_heads": 16,
143
+ "num_code_groups": 16,
144
+ "num_hidden_layers": 28,
145
+ "num_key_value_heads": 8,
146
+ "position_id_per_seconds": 13,
147
+ "rms_norm_eps": 1e-06,
148
+ "rope_scaling": {
149
+ "interleaved": true,
150
+ "mrope_section": [
151
+ 24,
152
+ 20,
153
+ 20
154
+ ],
155
+ "rope_type": "default",
156
+ "type": "default"
157
+ },
158
+ "rope_theta": 1000000,
159
+ "sliding_window": null,
160
+ "text_hidden_size": 2048,
161
+ "text_vocab_size": 151936,
162
+ "use_cache": true,
163
+ "use_sliding_window": false,
164
+ "vocab_size": 3072
165
+ },
166
+ "transformers_version": "4.57.3"
167
+ }
qwen3-tts/generation_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_sample": true,
3
+ "repetition_penalty": 1.05,
4
+ "temperature": 0.9,
5
+ "top_p": 1.0,
6
+ "top_k": 50,
7
+ "subtalker_dosample": true,
8
+ "subtalker_temperature": 0.9,
9
+ "subtalker_top_p": 1.0,
10
+ "subtalker_top_k": 50,
11
+ "max_new_tokens": 8192
12
+ }
qwen3-tts/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
qwen3-tts/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38fc7fc51c5e776e840414b6fd443962e9411b9654888fd7913e4da643cb857c
3
+ size 3857413744
qwen3-tts/preprocessor_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "padding_side": "left",
3
+ "padding_value": 0.0,
4
+ "processor_class": "Qwen3TTSProcessor",
5
+ "return_attention_mask": true
6
+ }
qwen3-tts/speech_tokenizer/config.json ADDED
@@ -0,0 +1,94 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3TTSTokenizerV2Model"
4
+ ],
5
+ "model_type": "qwen3_tts_tokenizer_12hz",
6
+ "encoder_valid_num_quantizers": 16,
7
+ "input_sample_rate": 24000,
8
+ "output_sample_rate": 24000,
9
+ "decode_upsample_rate": 1920,
10
+ "encode_downsample_rate": 1920,
11
+ "decoder_config": {
12
+ "attention_bias": false,
13
+ "attention_dropout": 0.0,
14
+ "latent_dim": 1024,
15
+ "codebook_dim": 512,
16
+ "codebook_size": 2048,
17
+ "decoder_dim": 1536,
18
+ "hidden_act": "silu",
19
+ "hidden_size": 512,
20
+ "intermediate_size": 1024,
21
+ "layer_scale_initial_scale": 0.01,
22
+ "max_position_embeddings": 8000,
23
+ "head_dim": 64,
24
+ "num_attention_heads": 16,
25
+ "num_hidden_layers": 8,
26
+ "num_key_value_heads": 16,
27
+ "num_quantizers": 16,
28
+ "num_semantic_quantizers": 1,
29
+ "rms_norm_eps": 1e-05,
30
+ "rope_theta": 10000,
31
+ "semantic_codebook_size": 4096,
32
+ "sliding_window": 72,
33
+ "upsample_rates": [
34
+ 8,
35
+ 5,
36
+ 4,
37
+ 3
38
+ ],
39
+ "upsampling_ratios": [
40
+ 2,
41
+ 2
42
+ ],
43
+ "vector_quantization_hidden_dimension": 512
44
+ },
45
+ "encoder_config": {
46
+ "_frame_rate": 12.5,
47
+ "attention_bias": false,
48
+ "attention_dropout": 0.0,
49
+ "audio_channels": 1,
50
+ "codebook_dim": 256,
51
+ "codebook_size": 2048,
52
+ "compress": 2,
53
+ "dilation_growth_rate": 2,
54
+ "dtype": "float32",
55
+ "head_dim": 64,
56
+ "hidden_act": "gelu",
57
+ "hidden_size": 512,
58
+ "initializer_range": 0.02,
59
+ "intermediate_size": 2048,
60
+ "kernel_size": 7,
61
+ "last_kernel_size": 3,
62
+ "layer_scale_initial_scale": 0.01,
63
+ "max_position_embeddings": 8000,
64
+ "norm_eps": 1e-05,
65
+ "normalize": false,
66
+ "num_attention_heads": 8,
67
+ "num_filters": 64,
68
+ "num_hidden_layers": 8,
69
+ "num_key_value_heads": 8,
70
+ "num_quantizers": 32,
71
+ "num_residual_layers": 1,
72
+ "num_semantic_quantizers": 1,
73
+ "pad_mode": "constant",
74
+ "residual_kernel_size": 3,
75
+ "rope_theta": 10000.0,
76
+ "sampling_rate": 24000,
77
+ "sliding_window": 250,
78
+ "transformers_version": "4.57.0.dev0",
79
+ "trim_right_ratio": 1.0,
80
+ "upsample_groups": 512,
81
+ "upsampling_ratios": [
82
+ 8,
83
+ 6,
84
+ 5,
85
+ 4
86
+ ],
87
+ "use_cache": false,
88
+ "use_causal_conv": true,
89
+ "use_conv_shortcut": false,
90
+ "use_streaming": false,
91
+ "vector_quantization_hidden_dimension": 256
92
+ },
93
+ "transformers_version": "4.57.3"
94
+ }
qwen3-tts/speech_tokenizer/configuration.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"framework": "pytorch", "task": "feature-extraction", "allow_remote": true}
qwen3-tts/speech_tokenizer/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:836b7b357f5ea43e889936a3709af68dfe3751881acefe4ecf0dbd30ba571258
3
+ size 682293092
qwen3-tts/speech_tokenizer/preprocessor_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length_s": null,
3
+ "feature_extractor_type": "EncodecFeatureExtractor",
4
+ "feature_size": 1,
5
+ "overlap": null,
6
+ "padding_side": "right",
7
+ "padding_value": 0.0,
8
+ "return_attention_mask": true,
9
+ "sampling_rate": 24000
10
+ }
qwen3-tts/tokenizer_config.json ADDED
@@ -0,0 +1,316 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_prefix_space": false,
4
+ "added_tokens_decoder": {
5
+ "151643": {
6
+ "content": "<|endoftext|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "151644": {
14
+ "content": "<|im_start|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "151645": {
22
+ "content": "<|im_end|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "151646": {
30
+ "content": "<|object_ref_start|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "151647": {
38
+ "content": "<|object_ref_end|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "151648": {
46
+ "content": "<|box_start|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "151649": {
54
+ "content": "<|box_end|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ },
61
+ "151650": {
62
+ "content": "<|quad_start|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": true
68
+ },
69
+ "151651": {
70
+ "content": "<|quad_end|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "151652": {
78
+ "content": "<|vision_start|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ },
85
+ "151653": {
86
+ "content": "<|vision_end|>",
87
+ "lstrip": false,
88
+ "normalized": false,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": true
92
+ },
93
+ "151654": {
94
+ "content": "<|vision_pad|>",
95
+ "lstrip": false,
96
+ "normalized": false,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": true
100
+ },
101
+ "151655": {
102
+ "content": "<|image_pad|>",
103
+ "lstrip": false,
104
+ "normalized": false,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": true
108
+ },
109
+ "151656": {
110
+ "content": "<|video_pad|>",
111
+ "lstrip": false,
112
+ "normalized": false,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": true
116
+ },
117
+ "151657": {
118
+ "content": "<tool_call>",
119
+ "lstrip": false,
120
+ "normalized": false,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": false
124
+ },
125
+ "151658": {
126
+ "content": "</tool_call>",
127
+ "lstrip": false,
128
+ "normalized": false,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": false
132
+ },
133
+ "151659": {
134
+ "content": "<|fim_prefix|>",
135
+ "lstrip": false,
136
+ "normalized": false,
137
+ "rstrip": false,
138
+ "single_word": false,
139
+ "special": false
140
+ },
141
+ "151660": {
142
+ "content": "<|fim_middle|>",
143
+ "lstrip": false,
144
+ "normalized": false,
145
+ "rstrip": false,
146
+ "single_word": false,
147
+ "special": false
148
+ },
149
+ "151661": {
150
+ "content": "<|fim_suffix|>",
151
+ "lstrip": false,
152
+ "normalized": false,
153
+ "rstrip": false,
154
+ "single_word": false,
155
+ "special": false
156
+ },
157
+ "151662": {
158
+ "content": "<|fim_pad|>",
159
+ "lstrip": false,
160
+ "normalized": false,
161
+ "rstrip": false,
162
+ "single_word": false,
163
+ "special": false
164
+ },
165
+ "151663": {
166
+ "content": "<|repo_name|>",
167
+ "lstrip": false,
168
+ "normalized": false,
169
+ "rstrip": false,
170
+ "single_word": false,
171
+ "special": false
172
+ },
173
+ "151664": {
174
+ "content": "<|file_sep|>",
175
+ "lstrip": false,
176
+ "normalized": false,
177
+ "rstrip": false,
178
+ "single_word": false,
179
+ "special": false
180
+ },
181
+ "151665": {
182
+ "content": "<tool_response>",
183
+ "lstrip": false,
184
+ "normalized": false,
185
+ "rstrip": false,
186
+ "single_word": false,
187
+ "special": false
188
+ },
189
+ "151666": {
190
+ "content": "</tool_response>",
191
+ "lstrip": false,
192
+ "normalized": false,
193
+ "rstrip": false,
194
+ "single_word": false,
195
+ "special": false
196
+ },
197
+ "151667": {
198
+ "content": "<think>",
199
+ "lstrip": false,
200
+ "normalized": false,
201
+ "rstrip": false,
202
+ "single_word": false,
203
+ "special": false
204
+ },
205
+ "151668": {
206
+ "content": "</think>",
207
+ "lstrip": false,
208
+ "normalized": false,
209
+ "rstrip": false,
210
+ "single_word": false,
211
+ "special": false
212
+ },
213
+ "151669": {
214
+ "content": "<|audio_start|>",
215
+ "lstrip": false,
216
+ "normalized": false,
217
+ "rstrip": false,
218
+ "single_word": false,
219
+ "special": true
220
+ },
221
+ "151670": {
222
+ "content": "<|audio_end|>",
223
+ "lstrip": false,
224
+ "normalized": false,
225
+ "rstrip": false,
226
+ "single_word": false,
227
+ "special": true
228
+ },
229
+ "151671": {
230
+ "content": "<tts_pad>",
231
+ "lstrip": false,
232
+ "normalized": false,
233
+ "rstrip": false,
234
+ "single_word": false,
235
+ "special": true
236
+ },
237
+ "151672": {
238
+ "content": "<tts_text_bos>",
239
+ "lstrip": false,
240
+ "normalized": false,
241
+ "rstrip": false,
242
+ "single_word": false,
243
+ "special": true
244
+ },
245
+ "151673": {
246
+ "content": "<tts_text_eod>",
247
+ "lstrip": false,
248
+ "normalized": false,
249
+ "rstrip": false,
250
+ "single_word": false,
251
+ "special": true
252
+ },
253
+ "151674": {
254
+ "content": "<tts_text_bos_single>",
255
+ "lstrip": false,
256
+ "normalized": false,
257
+ "rstrip": false,
258
+ "single_word": false,
259
+ "special": true
260
+ },
261
+ "151675": {
262
+ "content": "<|audio_pad|>",
263
+ "lstrip": false,
264
+ "normalized": false,
265
+ "rstrip": false,
266
+ "single_word": false,
267
+ "special": true
268
+ }
269
+ },
270
+ "additional_special_tokens": [
271
+ "<|im_start|>",
272
+ "<|im_end|>",
273
+ "<|object_ref_start|>",
274
+ "<|object_ref_end|>",
275
+ "<|box_start|>",
276
+ "<|box_end|>",
277
+ "<|quad_start|>",
278
+ "<|quad_end|>",
279
+ "<|vision_start|>",
280
+ "<|vision_end|>",
281
+ "<|vision_pad|>",
282
+ "<|image_pad|>",
283
+ "<|video_pad|>",
284
+ "<|audio_start|>",
285
+ "<|audio_end|>",
286
+ "<tts_pad>",
287
+ "<tts_text_bos>",
288
+ "<tts_text_bos_single>",
289
+ "<|audio_pad|>"
290
+ ],
291
+ "extra_special_tokens": {
292
+ "image_token": "<|image_pad|>",
293
+ "audio_token": "<|audio_pad|>",
294
+ "video_token": "<|video_pad|>",
295
+ "vision_bos_token": "<|vision_start|>",
296
+ "vision_eos_token": "<|vision_end|>",
297
+ "audio_bos_token": "<|audio_start|>",
298
+ "audio_eos_token": "<|audio_end|>"
299
+ },
300
+ "bos_token": null,
301
+ "clean_up_tokenization_spaces": false,
302
+ "eos_token": "<|im_end|>",
303
+ "errors": "replace",
304
+ "model_max_length": 131072,
305
+ "pad_token": "<|endoftext|>",
306
+ "split_special_tokens": false,
307
+ "tokenizer_class": "Qwen2Tokenizer",
308
+ "unk_token": null,
309
+ "image_token": "<|image_pad|>",
310
+ "audio_token": "<|audio_pad|>",
311
+ "video_token": "<|video_pad|>",
312
+ "vision_bos_token": "<|vision_start|>",
313
+ "vision_eos_token": "<|vision_end|>",
314
+ "audio_bos_token": "<|audio_start|>",
315
+ "audio_eos_token": "<|audio_end|>"
316
+ }
qwen3-tts/vocab.json ADDED
The diff for this file is too large to render. See raw diff