Connection Error

[jeonhuhuhu]

"I would appreciate it if you could kindly provide the necessary details and data associated with code download_checkpoints_and_tokenizers.py again.

I'm currently facing an issue preventing me from downloading the related information, and I need it to run the code successfully.

Thank you for your assistance."

[jeonhuhuhu]

"I would appreciate it if you could kindly provide the necessary details and data associated with code download_checkpoints_and_tokenizers.py again.

I'm currently facing an issue preventing me from downloading the related information, and I need it to run the code successfully.

Thank you for your assistance."

We manually downloaded and processed the data one by one.

[burchim]

Hi, what was the error or issue that you encountered ?

The demo notebook provide a demo on how to download checkpoints and evaluate the models: https://colab.research.google.com/github/burchim/AVEC/blob/master/demo.ipynb

Best,
Maxime

[jeonhuhuhu]

Hi, what was the error or issue that you encountered ?

The demo notebook provide a demo on how to download checkpoints and evaluate the models: https://colab.research.google.com/github/burchim/AVEC/blob/master/demo.ipynb

Best, Maxime

There was an issue with your Google Drive connection preventing outsiders from accessing it, but it has been resolved. For the second problem, can you share a detailed version of CUDA, Pytorch, torchaudio, torchvision?

[burchim]

Yes Google Drive can sometimes returns an error saying that the file was downloaded too many times in a short period.

Concerning the version of libraries, I was using PyTorch 1.12 with torchaudio 0.12 and torchvision 0.13.

I was also using cuda 11.2 for A100 GPU support.

[jeonhuhuhu]

Yes Google Drive can sometimes returns an error saying that the file was downloaded too many times in a short period.

Concerning the version of libraries, I was using PyTorch 1.12 with torchaudio 0.12 and torchvision 0.13.

I was also using cuda 11.2 for A100 GPU support.

Your installation process is perfectly well explained. However, if we repeatedly install ctcdecode, the following error keeps occurring and I cannot install it. Can you share the installation method?

subprocess.CalledProcessError: Command '['which', 'x86_64-conda_cos7-linux-gnu-c++']' returned non-zero exit status 1.
[end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for ctcdecode
Running setup.py clean for ctcdecode
Failed to build ctcdecode
ERROR: Could not build wheels for ctcdecode, which is required to install pyproject.toml-based projects

[jeonhuhuhu]

Yes Google Drive can sometimes returns an error saying that the file was downloaded too many times in a short period.
Concerning the version of libraries, I was using PyTorch 1.12 with torchaudio 0.12 and torchvision 0.13.
I was also using cuda 11.2 for A100 GPU support.

Your installation process is perfectly well explained. However, if we repeatedly install ctcdecode, the following error keeps occurring and I cannot install it. Can you share the installation method?

subprocess.CalledProcessError: Command '['which', 'x86_64-conda_cos7-linux-gnu-c++']' returned non-zero exit status 1. [end of output]

note: This error originates from a subprocess, and is likely not a problem with pip. ERROR: Failed building wheel for ctcdecode Running setup.py clean for ctcdecode Failed to build ctcdecode ERROR: Could not build wheels for ctcdecode, which is required to install pyproject.toml-based projects

"I have managed to succeed in the installation and have shared the installation method."
parlance/ctcdecode#229

[burchim]

Hi, thanks for your return.

Yes CTCdecode can sometimes fail to install when using specific PyTorch versions.
I know it is working well with PyTorch 2.0.0 and under.

Hope you could successfully evaluate the models and/or do some experiments with the repo.