'install torchaudio on AWS training container
What is the proper way to install torchaudio in AWS sagemaker GPU container? and why is it not pre-installed?
I tried pip install torchaudio but it tries to uninstall and install pytorch as well that breaks the whole container. I got segmentation fault error probably because it installed CPU based version of pytorch and torchaudio.
Solution 1:[1]
The fact that pip auto uninstall and installs Pytorch probably shows that you're installing a torchaudio version which is uncompatible with the pytorch version. Perhaps:
- Find a matching set of PyTorch and TorchAudio. Test that they install well locally.
- Make sure you find an AWS DLC container with such PyTorch version.
- pip install torchaudio specifying the exact version you need. Test this with Docker on your local machine, cloud not needed.
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|---|
| Solution 1 | Dharman |