mps device

[zachysaur]

(venv) F:\talker\liveportrait_talker-main>python inference.py --config_path config.yaml --source_path "F:\talker\liveportrait_talker-main\Face\1.png" --audio_path "F:\talker\liveportrait_talker-main\Audio\2.wav" --save_path "F:\talker\liveportrait_talker-main\Audio" Config File is loaded succesfully! Traceback (most recent call last): File "F:\talker\liveportrait_talker-main\inference.py", line 78, in main(args) File "F:\talker\liveportrait_talker-main\inference.py", line 23, in main preprocess = Preprocess(device=cfg.device, File "F:\talker\liveportrait_talker-main\src\modules\preprocess.py", line 28, in init self.sd_prep = SadTalkerPreprocess(device=device, File "F:\talker\liveportrait_talker-main\src\utils\preprocess\sadtalker_preprocess.py", line 20, in init self.detector = init_alignment_model('awing_fan', device=device, model_rootpath=model_path) File "F:\talker\liveportrait_talker-main\src\utils\preprocess\helper.py", line 149, in init_alignment_model model = model.to(device) File "F:\talker\liveportrait_talker-main\venv\lib\site-packages\torch\nn\modules\module.py", line 1173, in to return self._apply(convert) File "F:\talker\liveportrait_talker-main\venv\lib\site-packages\torch\nn\modules\module.py", line 779, in _apply module._apply(fn) File "F:\talker\liveportrait_talker-main\venv\lib\site-packages\torch\nn\modules\module.py", line 779, in _apply module._apply(fn) File "F:\talker\liveportrait_talker-main\venv\lib\site-packages\torch\nn\modules\module.py", line 804, in _apply param_applied = fn(param) File "F:\talker\liveportrait_talker-main\venv\lib\site-packages\torch\nn\modules\module.py", line 1159, in convert return t.to( RuntimeError: PyTorch is not linked with support for mps devices

(venv) F:\talker\liveportrait_talker-main>

[mkara44]

You need to change device type from config file (inference.device). If you are using gpu you have to set device as cuda, if not you have to set as cpu. In default, i set as mps which is for macbook.

[zachysaur]

Microsoft Windows [Version 10.0.19045.5011] (c) Microsoft Corporation. All rights reserved.

F:\talker\liveportrait_talker-main>venv\scripts\activate

(venv) F:\talker\liveportrait_talker-main>python inference.py --config_path config.yaml --source_path "F:\talker\liveportrait_talker-main\Face\1.png" --audio_path "F:\talker\liveportrait_talker-main\Audio\2.wav" --save_path "F:\talker\liveportrait_talker-main\Audio\1.mp4" Config File is loaded succesfully! F:\talker\liveportrait_talker-main\venv\lib\site-packages\torchvision\models_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead. warnings.warn( F:\talker\liveportrait_talker-main\venv\lib\site-packages\torchvision\models_utils.py:223: UserWarning: Arguments other than a weight enum or None for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing weights=None. warnings.warn(msg) Pipeline Objects are initialized! LLVM ERROR: Symbol not found: __svml_cosf8_ha

(venv) F:\talker\liveportrait_talker-main>

[mkara44]

Hi, I haven't seen this issue before. I checked library versions too, everything is seen okay and works fine. You can try this solution.

[zachysaur]

fixed it somehow i will make a fork and see if these changes needed

[zachysaur]

https://youtu.be/6aS-Ui37o-o lol u did models changes after i made video

[zachysaur]

there also two payyaml versions in requirements cause error

[mkara44]

Wow, thanks for yout support. I haven't seen this video sorry 😄. I also realized that head pose is not working in your video, I updated the pipeline. I guess I solved the issus.

Also, requirements.txt problem is solved too.

Thanks for your feedback too.

[zachysaur]

i am trying to make gradio of it i am not good at it but somehow i manage to make things work

[mkara44]

Gradio is on my pipeline too, but I don't know when can I complete it because of my work.