Sihan Chen comments

Results 31 comments of


                                            Sihan Chen

deploy sadtalker with trition by cuda optimization can see here. real time.

@foocker great work! 2.4s also includes face enhancing (gfpgan) time, right?

deploy sadtalker with trition by cuda optimization can see here. real time.

> > @foocker great work! 2.4s also includes face enhancing (gfpgan) time, right? > > no, the result is ok, so i removed it. Cool!

RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.

Hi @diabolo98 , this error lies in the facexlib dependencies, not in inference.py. Please in your `/usr/local/lib/python3.8/dist-packages/facexlib/alignment/__init__.py` , try edit the line ``` model.load_state_dict(torch.load(model_path)['state_dict'], strict=True) ``` to ``` model.load_state_dict(torch.load(model_path, map_location=device)['state_dict'],...

RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.

Hi @diabolo98 , I just noticed that you are doing the experiment on colab. I have not ever tested it on colab with **cpu only**. colab's cpus are limited. I...

RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.

int8 optimization itself does not depend on intel extension for pytorch here so it will not raise the above errors. But again, I've not tested it yet in colab environment....

RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.

Colab is truly limited in CPU because it tends to let you use GPU or TPU to do computing intensive work. My optimization is tested on Xeon Sapphire Rapids (check...

RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.

No problem, your feedbacks are welcome :)

Attempting to deserialize object on CUDA device

@SuperMaximus1984 , please check https://github.com/Spycsh/xtalker/issues/3

Hello, I came across your project and I'm interested to know if it has the capability to run in real-time and drive images to speak?

Thanks for interest. Currently XTalker is able to drive image to speak with about 10x speed compared to SadTalker. I have not integrated it into any real-time streaming system. However,...

MPS device compatibility

Thanks for info:) I do not have access to a Mac currently and the optimization here in my repo should only be used for intel xeon cpu. But good to...