[index-tts]安装v2的时候,命令uvsync --all-extras执行到配置deepspeed时出现错误

2025-10-28 35 views
1

x Failed to builddeepspeed==0.17.1 |-> The build backend returned an error -> Call to setuptools.build_meta:__legacy__.build_wheel failed (exit code: 1)

  [stdout]
  [2025-09-09 19:32:28,936] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto
  detect)
  [2025-09-09 19:32:29,026] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto
  detect)
  [2025-09-09 19:32:31,529] [INFO] [logging.py:107:log_dist] [Rank -1] [TorchCheckpointEngine] Initialized with
  serialization = False
  DS_BUILD_OPS=1
   [WARNING]  Skip pre-compile of incompatible async_io; One can disable async_io with DS_BUILD_AIO=0
   [WARNING]  Skip pre-compile of incompatible evoformer_attn; One can disable evoformer_attn with
  DS_BUILD_EVOFORMER_ATTN=0
   [WARNING]  Skip pre-compile of incompatible fp_quantizer; One can disable fp_quantizer with
  DS_BUILD_FP_QUANTIZER=0
   [WARNING]  Skip pre-compile of incompatible gds; One can disable gds with DS_BUILD_GDS=0
   [WARNING]  Filtered compute capabilities [['6', '0'], ['6', '1'], ['7', '0']]

  [stderr]
  W0909 19:32:30.260000 28180 Lib\site-packages\torch\distributed\elastic\multiprocessing\redirects.py:29] NOTE:
  Redirects are currently not supported in Windows or MacOs.
  Traceback (most recent call last):
    File "<string>", line 11, in <module>
    File "D:\MyLab\IndexTTS\index-tts\.venv\lib\site-packages\setuptools\build_meta.py", line 432, in build_wheel
      return _build(['bdist_wheel'])
    File "D:\MyLab\IndexTTS\index-tts\.venv\lib\site-packages\setuptools\build_meta.py", line 423, in _build
      return self._build_with_temp_dir(
    File "D:\MyLab\IndexTTS\index-tts\.venv\lib\site-packages\setuptools\build_meta.py", line 404, in
  _build_with_temp_dir
      self.run_setup()
    File "D:\MyLab\IndexTTS\index-tts\.venv\lib\site-packages\setuptools\build_meta.py", line 512, in run_setup
      super().run_setup(setup_script=setup_script)
    File "D:\MyLab\IndexTTS\index-tts\.venv\lib\site-packages\setuptools\build_meta.py", line 317, in run_setup
      exec(code, locals())
    File "<string>", line 201, in <module>
    File
  "C:\Users\WangQ\AppData\Local\uv\cache\sdists-v9\pypi\deepspeed\0.17.1\R5RYAY0102coQzee_--Dk\src\op_builder\builder.py",
  line 730, in builder
      extra_link_args=self.strip_empty_entries(self.extra_ldflags()))
    File
  "C:\Users\WangQ\AppData\Local\uv\cache\sdists-v9\pypi\deepspeed\0.17.1\R5RYAY0102coQzee_--Dk\src\op_builder\inference_cutlass_builder.py",
  line 74, in extra_ldflags
      import dskernels
  ModuleNotFoundError: No module named 'dskernels'

  hint: This error likely indicates that `deepspeed@0.17.1` depends on `dskernels`, but doesn't declare
  it as a build dependency. If `deepspeed` is a first-party package, consider adding `dskernels` to its
  `build-system.requires`. Otherwise, either add it to your `pyproject.toml` under:

  [tool.uv.extra-build-dependencies]
  deepspeed = ["dskernels"]

  or `uv pip install dskernels` into the environment and re-run with `--no-build-isolation`.

help: deepspeed (v0.17.1) was included because indextts (v2.0.0) depends on deepspeed

`

系统是WIN 11

回答

0

同,我改了CUDA12.4,也报错

1

DeepSpeed 软件包现已转换为可选依赖项。安装文档也已更新。

您可以通过运行此同步命令来安装没有 DeepSpeed 的 IndexTTS:

uv sync --extra webui

这会显示 Web UI,但会跳过 DeepSpeed 的安装。这对于 Windows 用户来说可能比较实用,因为 Windows 是安装 DeepSpeed 最困难的平台。在 Linux 上,它似乎适用于所有用户。

但 DeepSpeed 通常比普通推理速度慢,所以你不必担心错过它。请参阅以下 RTX 3090 和 5090 基准测试:https://github.com/index-tts/index-tts/issues/326#issuecomment-3321924827