MuseTalk模型 v1.0.0，运行不了

升级最新版本试一下 1.1.0

@MZ: 升级到1.1.0了，运行日志报错如下：

194[I] 2026-05-09 11:04:29 - server starting version=0.0.1 buildTime=2026-01-28 18:54:40 debug=false

195[I] 2026-05-09 11:04:30 - start command mode

196[I] 2026-05-09 11:04:30 - Setting offline envs for model downloading

197[I] 2026-05-09 11:04:30 - LIB.InitEnvFromArgs - {}

198[I] 2026-05-09 11:04:30 - LIB.Start

199[I] 2026-05-09 11:04:31 - LIB.AigcPanelServer - {"device":"cuda","name":"MuseTalk","version":"1.1.0","mode":{"type":"watch","watchDelay":60},"root":"D:\\AIGCPANEL\\mold\\musetalk1.1","platformName":"win","platformArch":"x86"}

200[I] 2026-05-09 11:04:31 - LIB.Argv - ["D:\\AIGCPANEL\\mold\\musetalk1.1\\aigcpanelrun.py","C:\\Users\\Administrator\\AppData\\Roaming\\aigcpanel\\data\\temp\\file_1778295869775_retzB7setbHnC3oK2AI45qStwErw7Wqj.json"]

201[I] 2026-05-09 11:04:31 - LIB.Config - {"id":"VideoGen_39","mode":"local","modelConfig":{"type":"videoGen","param":{},"video":"C:\\Users\\Administrator\\AppData\\Roaming\\aigcpanel\\data\\hub\\file\\20260429\\2043_46_KkiitoBom4JKh0aAPLg4egsyN1DFMsMO.mp4","audio":"C:\\Users\\Administrator\\Desktop\\01.mp3"},"setting":{"gpu":"0"}}

202[I] 2026-05-09 11:04:31 - LIB.Set - CUDA_VISIBLE_DEVICES to 0

203D:\AIGCPANEL\mold\musetalk1.1\_aienv\lib\site-packages\torch\cuda\__init__.py:235: UserWarning:

204NVIDIA GeForce RTX 5060 Ti with CUDA capability sm_120 is not compatible with the current PyTorch installation.

205The current PyTorch install supports CUDA capabilities sm_50 sm_60 sm_61 sm_70 sm_75 sm_80 sm_86 sm_90.

206If you want to use the NVIDIA GeForce RTX 5060 Ti GPU with PyTorch, please check the instructions at https://pytorch.org/get-started/locally/

207

208warnings.warn(

209[I] 2026-05-09 11:04:31 - LIB.Info - {"Device":"cuda","DeviceName":"NVIDIA GeForce RTX 5060 Ti","DeviceMemorySize":8,"CudaVersion":"13.2"}

210

D:\AIGCPANEL\mold\musetalk1.1\_aienv\lib\site-packages\mmengine\optim\optimizer\zero_optimizer.py:11: DeprecationWarning: `TorchScript` support for functional optimizers is deprecated and will be removed in a future PyTorch release. Consider using the `torch.compile` optimizer instead.

211from torch.distributed.optim import \

212[I] 2026-05-09 11:04:32 - MuseTalk.Info - Loading models...

213Loads checkpoint by local backend from path: D:\AIGCPANEL\mold\musetalk1.1\_cache/huggingface/hub/models--yzd-v--DWPose\snapshots\1a7144101628d69ee7a3768d1ee3a094070dc388\dw-ll_ucoco_384.pth

214cuda start

215An error occurred while trying to fetch D:\AIGCPANEL\mold\musetalk1.1\_cache/huggingface/hub/models--stabilityai--sd-vae-ft-mse\snapshots\31f26fdeee1355a5c34592e401dd41e45d25a493: Error no file named diffusion_pytorch_model.safetensors found in directory D:\AIGCPANEL\mold\musetalk1.1\_cache/huggingface/hub/models--stabilityai--sd-vae-ft-mse\snapshots\31f26fdeee1355a5c34592e401dd41e45d25a493.

216Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.

217load unet model from D:\AIGCPANEL\mold\musetalk1.1\_cache/huggingface/hub/models--TMElyralab--MuseTalk\snapshots\3ef28bc5cff08c90ad8178a25f1b570cd800170f\musetalkV15/unet.pth

218[I] 2026-05-09 11:04:38 - UNCAUGHT.EXCEPTION - {"type":"<class 'RuntimeError'>","value":"CUDA error: no kernel image is available for execution on the device\nCUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.\nFor debugging consider passing CUDA_LAUNCH_BLOCKING=1\nCompile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.\n"}219

Traceback (most recent call last):

220

File "D:\AIGCPANEL\mold\musetalk1.1\_aienv\lib\runpy.py", line 196, in _run_module_as_main

221return _run_code(code, main_globals, None,

222File "D:\AIGCPANEL\mold\musetalk1.1\_aienv\lib\runpy.py", line 86, in _run_code

223exec(code, run_globals)

224File "D:\AIGCPANEL\mold\musetalk1.1\aigcpanelrun.py", line 2, in <module>

225aigcpanelapp.run()

226File "aigcpanelapp.py", line 148, in aigcpanelapp.run

227File "D:\AIGCPANEL\mold\musetalk1.1\_aienv\lib\site-packages\torch\nn\modules\module.py", line 1165, in half

228return self._apply(lambda t: t.half() if t.is_floating_point() else t)

229File "D:\AIGCPANEL\mold\musetalk1.1\_aienv\lib\site-packages\torch\nn\modules\module.py", line 900, in _apply

230module._apply(fn)

231File "D:\AIGCPANEL\mold\musetalk1.1\_aienv\lib\site-packages\torch\nn\modules\module.py", line 900, in _apply

232module._apply(fn)

233File "D:\AIGCPANEL\mold\musetalk1.1\_aienv\lib\site-packages\torch\nn\modules\module.py", line 927, in _apply

234param_applied = fn(param)

235File "D:\AIGCPANEL\mold\musetalk1.1\_aienv\lib\site-packages\torch\nn\modules\module.py", line 1165, in <lambda>

236return self._apply(lambda t: t.half() if t.is_floating_point() else t)

237RuntimeError: CUDA error: no kernel image is available for execution on the device

238

CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.

239For debugging consider passing CUDA_LAUNCH_BLOCKING=1

240Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

241

242[E] 2026-05-09 11:04:39 - Error:exit status 1

~~~~~~~~~~~~~~报错日志~~~~~~~~~~~~~~~~~~~~

硬件和软件条件：安装了Python 3.10，CPU是AMD 9950x，显卡是5060Ti

这个是“你的显卡太新了（RTX 5060 Ti），而你当前程序内置的 PyTorch 环境太旧了，无法识别这块显卡。”这个项目依赖的库默认没支持这么新，后面我们看一下能不能升级一下。

好吧