Installing on a 5090

[bloxea]

Hi all - Thanks very much in advance for your help. I have a new machine with a RTX 5090 and am getting this error on install.

UserWarning: NVIDIA GeForce RTX 5090 with CUDA capability sm_120 is not compatible with the current PyTorch installation. The current PyTorch install supports CUDA capabilities sm_50 sm_60 sm_61 sm_70 sm_75 sm_80 sm_86 sm_90.

Is there a way for me to modify the installation or fix the process at this point to proceed? I'm not a developer so am perhaps more lost than most on what steps I might take to either upgrade PyTorch or something else that would make AllTalk TTS work.

Sounds like this perhaps the same issue but not sure if the solution is obvious already to those who know more than me: #551

[nelwyn2]

You have to enter the environment for AllTalk and uninstall torch, torchaudio and torchvision and then install torch 2.7.0 along with the latest torchaudio and torchvision. You also need CUDA 12.8. Once those are updated it will work. This is the case for any other AI tools such as Stable Diffusion. I got the same error and did those steps to make it work again.

• Enter the Conda environment for AllTalk
• pip uninstall torch torchvision torchaudio
• pip install torch==2.7.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

You could also choose to use another engine such as Piper which doesn't require that stuff and it should just work as is.

[b333kay]

I switched to RTX 5060 Ti from RTX 3060 and I've found a way to install the new torch with cu128 but now getting below error, been trying to debug this for couple days now, any suggestions??

Tried clean git clone as well but no luck. Only way I'm able to get the new PyTorch and Cuda detected is if I go into
\alltalk_tts\alltalk_environment\env\Scripts
then I run the following outside my global PATH

./pip.exe uninstall torch torchvision torchaudio
./pip.exe  install torch==2.7.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

PS D:\My Directory\alltalk_tts> .\start_alltalk.bat
[AllTalk Startup]     _    _ _ _____     _ _       _____ _____ ____
[AllTalk Startup]    / \  | | |_   _|_ _| | | __  |_   _|_   _/ ___|
[AllTalk Startup]   / _ \ | | | | |/ _` | | |/ /    | |   | | \___ \
[AllTalk Startup]  / ___ \| | | | | (_| | |   <     | |   | |  ___) |
[AllTalk Startup] /_/   \_\_|_| |_|\__,_|_|_|\_\    |_|   |_| |____/
[AllTalk Startup]
[AllTalk Startup] Config file check      : No Updates required
D:\My Directory\alltalk_tts\alltalk_environment\env\Lib\site-packages\deepspeed\runtime\zero\linear.py:47: FutureWarning: `torch.cuda.amp.custom_fwd(args...)` is deprecated. Please use `torch.amp.custom_fwd(args..., device_type='cuda')` instead.
  @autocast_custom_fwd
D:\My Directory\alltalk_tts\alltalk_environment\env\Lib\site-packages\deepspeed\runtime\zero\linear.py:66: FutureWarning: `torch.cuda.amp.custom_bwd(args...)` is deprecated. Please use `torch.amp.custom_bwd(args..., device_type='cuda')` instead.
  @autocast_custom_bwd
[AllTalk Startup] AllTalk startup Mode   : Standalone mode
[AllTalk Startup] WAV file deletion      : Disabled
[AllTalk Startup] DeepSpeed version      : Not Detected
[AllTalk Startup] Model is available     : Checking
[AllTalk Startup] Model is available     : Checked
[AllTalk Startup] Current Python Version : 3.11.13
[AllTalk Startup] Current PyTorch Version: 2.7.0+cu128
[AllTalk Startup] Current CUDA Version   : 12.8
[AllTalk Startup] Current TTS Version    : 0.22.0
[AllTalk Startup] Current TTS Version is : Up to date
[AllTalk Startup] AllTalk Github updated : 19th March 2025 at 20:50
[AllTalk Startup] TTS Subprocess         : Starting up
[AllTalk Startup]
[AllTalk Startup] AllTalk Settings & Documentation: http://127.0.0.1:7851
[AllTalk Startup]
D:\My Directory\alltalk_tts\alltalk_environment\env\Lib\site-packages\deepspeed\runtime\zero\linear.py:47: FutureWarning: `torch.cuda.amp.custom_fwd(args...)` is deprecated. Please use `torch.amp.custom_fwd(args..., device_type='cuda')` instead.
  @autocast_custom_fwd
D:\My Directory\alltalk_tts\alltalk_environment\env\Lib\site-packages\deepspeed\runtime\zero\linear.py:66: FutureWarning: `torch.cuda.amp.custom_bwd(args...)` is deprecated. Please use `torch.amp.custom_bwd(args..., device_type='cuda')` instead.
  @autocast_custom_bwd
[AllTalk Model] XTTSv2 Local Loading xttsv2_2.0.2 into cuda
ERROR:    Traceback (most recent call last):
  File "D:\My Directory\alltalk_tts\alltalk_environment\env\Lib\site-packages\starlette\routing.py", line 734, in lifespan
    async with self.lifespan_context(app) as maybe_state:
  File "D:\My Directory\alltalk_tts\alltalk_environment\env\Lib\contextlib.py", line 210, in __aenter__
    return await anext(self.gen)
           ^^^^^^^^^^^^^^^^^^^^^
  File "D:\My Directory\alltalk_tts\tts_server.py", line 127, in startup_shutdown
    await setup()
  File "D:\My Directory\alltalk_tts\tts_server.py", line 172, in setup
    model = await xtts_manual_load_model()
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\My Directory\alltalk_tts\tts_server.py", line 244, in xtts_manual_load_model
    model.load_checkpoint(
  File "D:\My Directory\alltalk_tts\alltalk_environment\env\Lib\site-packages\TTS\tts\models\xtts.py", line 771, in load_checkpoint
    checkpoint = self.get_compatible_checkpoint_state_dict(model_path)
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\My Directory\alltalk_tts\alltalk_environment\env\Lib\site-packages\TTS\tts\models\xtts.py", line 714, in get_compatible_checkpoint_state_dict
    checkpoint = load_fsspec(model_path, map_location=torch.device("cpu"))["model"]
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\My Directory\alltalk_tts\alltalk_environment\env\Lib\site-packages\TTS\utils\io.py", line 54, in load_fsspec
    return torch.load(f, map_location=map_location, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\My Directory\alltalk_tts\alltalk_environment\env\Lib\site-packages\torch\serialization.py", line 1524, in load
    raise pickle.UnpicklingError(_get_wo_message(str(e))) from None
_pickle.UnpicklingError: Weights only load failed. This file can still be loaded, to do so you have two options, do those steps only if you trust the source of the checkpoint.
        (1) In PyTorch 2.6, we changed the default value of the `weights_only` argument in `torch.load` from `False` to `True`. Re-running `torch.load` with `weights_only` set to `False` will likely succeed, but it can result in arbitrary code execution. Do it only if you got the file from a trusted source.
        (2) Alternatively, to load with `weights_only=True` please check the recommended steps in the following error message.
        WeightsUnpickler error: Unsupported global: GLOBAL TTS.tts.configs.xtts_config.XttsConfig was not an allowed global by default. Please use `torch.serialization.add_safe_globals([TTS.tts.configs.xtts_config.XttsConfig])` or the `torch.serialization.safe_globals([TTS.tts.configs.xtts_config.XttsConfig])` context manager to allowlist this global if you trust this class/function.

Check the documentation of torch.load to learn more about types accepted by default with weights_only https://pytorch.org/docs/stable/generated/torch.load.html.

ERROR:    Application startup failed. Exiting.