xinference 1.0.0__py3-none-any.whl → 1.1.0__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.

This version of xinference might be problematic. Click here for more details.

Files changed (94) hide show
  1. xinference/_compat.py +22 -2
  2. xinference/_version.py +3 -3
  3. xinference/api/restful_api.py +91 -6
  4. xinference/client/restful/restful_client.py +39 -0
  5. xinference/core/model.py +41 -13
  6. xinference/deploy/cmdline.py +3 -1
  7. xinference/deploy/test/test_cmdline.py +56 -0
  8. xinference/isolation.py +24 -0
  9. xinference/model/audio/__init__.py +12 -0
  10. xinference/model/audio/core.py +26 -4
  11. xinference/model/audio/f5tts.py +195 -0
  12. xinference/model/audio/fish_speech.py +71 -35
  13. xinference/model/audio/model_spec.json +88 -0
  14. xinference/model/audio/model_spec_modelscope.json +9 -0
  15. xinference/model/audio/whisper_mlx.py +208 -0
  16. xinference/model/embedding/core.py +322 -6
  17. xinference/model/embedding/model_spec.json +8 -1
  18. xinference/model/embedding/model_spec_modelscope.json +9 -1
  19. xinference/model/llm/__init__.py +4 -2
  20. xinference/model/llm/llm_family.json +479 -53
  21. xinference/model/llm/llm_family_modelscope.json +423 -17
  22. xinference/model/llm/mlx/core.py +230 -50
  23. xinference/model/llm/sglang/core.py +2 -0
  24. xinference/model/llm/transformers/chatglm.py +9 -5
  25. xinference/model/llm/transformers/core.py +1 -0
  26. xinference/model/llm/transformers/glm_edge_v.py +230 -0
  27. xinference/model/llm/transformers/utils.py +16 -8
  28. xinference/model/llm/utils.py +23 -1
  29. xinference/model/llm/vllm/core.py +89 -2
  30. xinference/thirdparty/f5_tts/__init__.py +0 -0
  31. xinference/thirdparty/f5_tts/api.py +166 -0
  32. xinference/thirdparty/f5_tts/configs/E2TTS_Base_train.yaml +44 -0
  33. xinference/thirdparty/f5_tts/configs/E2TTS_Small_train.yaml +44 -0
  34. xinference/thirdparty/f5_tts/configs/F5TTS_Base_train.yaml +46 -0
  35. xinference/thirdparty/f5_tts/configs/F5TTS_Small_train.yaml +46 -0
  36. xinference/thirdparty/f5_tts/eval/README.md +49 -0
  37. xinference/thirdparty/f5_tts/eval/ecapa_tdnn.py +330 -0
  38. xinference/thirdparty/f5_tts/eval/eval_infer_batch.py +207 -0
  39. xinference/thirdparty/f5_tts/eval/eval_infer_batch.sh +13 -0
  40. xinference/thirdparty/f5_tts/eval/eval_librispeech_test_clean.py +84 -0
  41. xinference/thirdparty/f5_tts/eval/eval_seedtts_testset.py +84 -0
  42. xinference/thirdparty/f5_tts/eval/utils_eval.py +405 -0
  43. xinference/thirdparty/f5_tts/infer/README.md +191 -0
  44. xinference/thirdparty/f5_tts/infer/SHARED.md +74 -0
  45. xinference/thirdparty/f5_tts/infer/examples/basic/basic.toml +11 -0
  46. xinference/thirdparty/f5_tts/infer/examples/basic/basic_ref_en.wav +0 -0
  47. xinference/thirdparty/f5_tts/infer/examples/basic/basic_ref_zh.wav +0 -0
  48. xinference/thirdparty/f5_tts/infer/examples/multi/country.flac +0 -0
  49. xinference/thirdparty/f5_tts/infer/examples/multi/main.flac +0 -0
  50. xinference/thirdparty/f5_tts/infer/examples/multi/story.toml +19 -0
  51. xinference/thirdparty/f5_tts/infer/examples/multi/story.txt +1 -0
  52. xinference/thirdparty/f5_tts/infer/examples/multi/town.flac +0 -0
  53. xinference/thirdparty/f5_tts/infer/examples/vocab.txt +2545 -0
  54. xinference/thirdparty/f5_tts/infer/infer_cli.py +226 -0
  55. xinference/thirdparty/f5_tts/infer/infer_gradio.py +851 -0
  56. xinference/thirdparty/f5_tts/infer/speech_edit.py +193 -0
  57. xinference/thirdparty/f5_tts/infer/utils_infer.py +538 -0
  58. xinference/thirdparty/f5_tts/model/__init__.py +10 -0
  59. xinference/thirdparty/f5_tts/model/backbones/README.md +20 -0
  60. xinference/thirdparty/f5_tts/model/backbones/dit.py +163 -0
  61. xinference/thirdparty/f5_tts/model/backbones/mmdit.py +146 -0
  62. xinference/thirdparty/f5_tts/model/backbones/unett.py +219 -0
  63. xinference/thirdparty/f5_tts/model/cfm.py +285 -0
  64. xinference/thirdparty/f5_tts/model/dataset.py +319 -0
  65. xinference/thirdparty/f5_tts/model/modules.py +658 -0
  66. xinference/thirdparty/f5_tts/model/trainer.py +366 -0
  67. xinference/thirdparty/f5_tts/model/utils.py +185 -0
  68. xinference/thirdparty/f5_tts/scripts/count_max_epoch.py +33 -0
  69. xinference/thirdparty/f5_tts/scripts/count_params_gflops.py +39 -0
  70. xinference/thirdparty/f5_tts/socket_server.py +159 -0
  71. xinference/thirdparty/f5_tts/train/README.md +77 -0
  72. xinference/thirdparty/f5_tts/train/datasets/prepare_csv_wavs.py +139 -0
  73. xinference/thirdparty/f5_tts/train/datasets/prepare_emilia.py +230 -0
  74. xinference/thirdparty/f5_tts/train/datasets/prepare_libritts.py +92 -0
  75. xinference/thirdparty/f5_tts/train/datasets/prepare_ljspeech.py +65 -0
  76. xinference/thirdparty/f5_tts/train/datasets/prepare_wenetspeech4tts.py +125 -0
  77. xinference/thirdparty/f5_tts/train/finetune_cli.py +174 -0
  78. xinference/thirdparty/f5_tts/train/finetune_gradio.py +1846 -0
  79. xinference/thirdparty/f5_tts/train/train.py +75 -0
  80. xinference/types.py +2 -1
  81. xinference/web/ui/build/asset-manifest.json +3 -3
  82. xinference/web/ui/build/index.html +1 -1
  83. xinference/web/ui/build/static/js/{main.2f269bb3.js → main.4eb4ee80.js} +3 -3
  84. xinference/web/ui/build/static/js/main.4eb4ee80.js.map +1 -0
  85. xinference/web/ui/node_modules/.cache/babel-loader/8c5eeb02f772d02cbe8b89c05428d0dd41a97866f75f7dc1c2164a67f5a1cf98.json +1 -0
  86. {xinference-1.0.0.dist-info → xinference-1.1.0.dist-info}/METADATA +39 -18
  87. {xinference-1.0.0.dist-info → xinference-1.1.0.dist-info}/RECORD +92 -39
  88. {xinference-1.0.0.dist-info → xinference-1.1.0.dist-info}/WHEEL +1 -1
  89. xinference/web/ui/build/static/js/main.2f269bb3.js.map +0 -1
  90. xinference/web/ui/node_modules/.cache/babel-loader/bd6ad8159341315a1764c397621a560809f7eb7219ab5174c801fca7e969d943.json +0 -1
  91. /xinference/web/ui/build/static/js/{main.2f269bb3.js.LICENSE.txt → main.4eb4ee80.js.LICENSE.txt} +0 -0
  92. {xinference-1.0.0.dist-info → xinference-1.1.0.dist-info}/LICENSE +0 -0
  93. {xinference-1.0.0.dist-info → xinference-1.1.0.dist-info}/entry_points.txt +0 -0
  94. {xinference-1.0.0.dist-info → xinference-1.1.0.dist-info}/top_level.txt +0 -0
xinference/thirdparty/f5_tts/train/train.py ADDED
@@ -0,0 +1,75 @@
1
+ # training script.
2
+
3
+ import os
4
+ from importlib.resources import files
5
+
6
+ import hydra
7
+
8
+ from f5_tts.model import CFM, DiT, Trainer, UNetT
9
+ from f5_tts.model.dataset import load_dataset
10
+ from f5_tts.model.utils import get_tokenizer
11
+
12
+ os.chdir(str(files("f5_tts").joinpath("../.."))) # change working directory to root of project (local editable)
13
+
14
+
15
+ @hydra.main(version_base="1.3", config_path=str(files("f5_tts").joinpath("configs")), config_name=None)
16
+ def main(cfg):
17
+ tokenizer = cfg.model.tokenizer
18
+ mel_spec_type = cfg.model.mel_spec.mel_spec_type
19
+ exp_name = f"{cfg.model.name}_{mel_spec_type}_{cfg.model.tokenizer}_{cfg.datasets.name}"
20
+
21
+ # set text tokenizer
22
+ if tokenizer != "custom":
23
+ tokenizer_path = cfg.datasets.name
24
+ else:
25
+ tokenizer_path = cfg.model.tokenizer_path
26
+ vocab_char_map, vocab_size = get_tokenizer(tokenizer_path, tokenizer)
27
+
28
+ # set model
29
+ if "F5TTS" in cfg.model.name:
30
+ model_cls = DiT
31
+ elif "E2TTS" in cfg.model.name:
32
+ model_cls = UNetT
33
+ wandb_resume_id = None
34
+
35
+ model = CFM(
36
+ transformer=model_cls(**cfg.model.arch, text_num_embeds=vocab_size, mel_dim=cfg.model.mel_spec.n_mel_channels),
37
+ mel_spec_kwargs=cfg.model.mel_spec,
38
+ vocab_char_map=vocab_char_map,
39
+ )
40
+
41
+ # init trainer
42
+ trainer = Trainer(
43
+ model,
44
+ epochs=cfg.optim.epochs,
45
+ learning_rate=cfg.optim.learning_rate,
46
+ num_warmup_updates=cfg.optim.num_warmup_updates,
47
+ save_per_updates=cfg.ckpts.save_per_updates,
48
+ checkpoint_path=str(files("f5_tts").joinpath(f"../../{cfg.ckpts.save_dir}")),
49
+ batch_size=cfg.datasets.batch_size_per_gpu,
50
+ batch_size_type=cfg.datasets.batch_size_type,
51
+ max_samples=cfg.datasets.max_samples,
52
+ grad_accumulation_steps=cfg.optim.grad_accumulation_steps,
53
+ max_grad_norm=cfg.optim.max_grad_norm,
54
+ logger=cfg.ckpts.logger,
55
+ wandb_project="CFM-TTS",
56
+ wandb_run_name=exp_name,
57
+ wandb_resume_id=wandb_resume_id,
58
+ last_per_steps=cfg.ckpts.last_per_steps,
59
+ log_samples=True,
60
+ bnb_optimizer=cfg.optim.bnb_optimizer,
61
+ mel_spec_type=mel_spec_type,
62
+ is_local_vocoder=cfg.model.vocoder.is_local,
63
+ local_vocoder_path=cfg.model.vocoder.local_path,
64
+ )
65
+
66
+ train_dataset = load_dataset(cfg.datasets.name, tokenizer, mel_spec_kwargs=cfg.model.mel_spec)
67
+ trainer.train(
68
+ train_dataset,
69
+ num_workers=cfg.datasets.num_workers,
70
+ resumable_with_seed=666, # seed for shuffling dataset
71
+ )
72
+
73
+
74
+ if __name__ == "__main__":
75
+ main()
xinference/types.py CHANGED
@@ -71,7 +71,8 @@ class EmbeddingUsage(TypedDict):
71
71
  class EmbeddingData(TypedDict):
72
72
  index: int
73
73
  object: str
74
- embedding: List[float]
74
+ # support sparse embedding
75
+ embedding: Union[List[float], Dict[str, float]]
75
76
 
76
77
 
77
78
  class Embedding(TypedDict):
xinference/web/ui/build/asset-manifest.json CHANGED
@@ -1,14 +1,14 @@
1
1
  {
2
2
  "files": {
3
3
  "main.css": "./static/css/main.5061c4c3.css",
4
- "main.js": "./static/js/main.2f269bb3.js",
4
+ "main.js": "./static/js/main.4eb4ee80.js",
5
5
  "static/media/icon.webp": "./static/media/icon.4603d52c63041e5dfbfd.webp",
6
6
  "index.html": "./index.html",
7
7
  "main.5061c4c3.css.map": "./static/css/main.5061c4c3.css.map",
8
- "main.2f269bb3.js.map": "./static/js/main.2f269bb3.js.map"
8
+ "main.4eb4ee80.js.map": "./static/js/main.4eb4ee80.js.map"
9
9
  },
10
10
  "entrypoints": [
11
11
  "static/css/main.5061c4c3.css",
12
- "static/js/main.2f269bb3.js"
12
+ "static/js/main.4eb4ee80.js"
13
13
  ]
14
14
  }
xinference/web/ui/build/index.html CHANGED
@@ -1 +1 @@
1
- <!doctype html><html lang="en"><head><meta charset="utf-8"/><link rel="icon" href="./favicon.svg"/><meta name="viewport" content="width=device-width,initial-scale=1"/><meta name="theme-color" content="#000000"/><meta name="description" content="Web site created using create-react-app"/><link rel="apple-touch-icon" href="./logo192.png"/><link rel="manifest" href="./manifest.json"/><title>Xinference</title><script defer="defer" src="./static/js/main.2f269bb3.js"></script><link href="./static/css/main.5061c4c3.css" rel="stylesheet"></head><body><noscript>You need to enable JavaScript to run this app.</noscript><div id="root"></div></body></html>
1
+ <!doctype html><html lang="en"><head><meta charset="utf-8"/><link rel="icon" href="./favicon.svg"/><meta name="viewport" content="width=device-width,initial-scale=1"/><meta name="theme-color" content="#000000"/><meta name="description" content="Web site created using create-react-app"/><link rel="apple-touch-icon" href="./logo192.png"/><link rel="manifest" href="./manifest.json"/><title>Xinference</title><script defer="defer" src="./static/js/main.4eb4ee80.js"></script><link href="./static/css/main.5061c4c3.css" rel="stylesheet"></head><body><noscript>You need to enable JavaScript to run this app.</noscript><div id="root"></div></body></html>