[Training] Onnxruntime-training 1.18.0 for windows not available about onnxruntime HOT 4 OPEN

It is because our storage was full. pypi/support#4085

from onnxruntime.

Oh, I might be wrong. We have training CPU packages and GPU packages.

1. https://pypi.org/project/onnxruntime-training/#files
2. https://pypi.org/project/onnxruntime-training-cpu/#files

The second one supports Windows and macOS. The first one only has Linux packages, and currently we do not have a plan to publish Windows GPU packages for training.

@wschin, is it right?

from onnxruntime.

@snnn is correct. The current setup is as follows:

• Large model training - targets linux gpu scenarios. Windows packages are not made available. Installation command:
  • pip install onnxruntime-training
• On-device training - targets linux, macos, windows cpu. Installation command:
  • pip install -i https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/ORT/pypi/simple/ onnxruntime-training-cpu

Note that the large model training python package for gpu contains all the pieces for on-device training. So, the large model training linux gpu package can also be used for on-device training.
But we do not publish the windows gpu python package.

The getting started table has all the installation information: https://onnxruntime.ai/getting-started

https://onnxruntime.ai/getting-started

from onnxruntime.

@baijumeswani
@snnn
The existing Jupiter notebook for onnx runtime training on device..masked nlp ..not updated for windows onnxrunetime training v 1.18.0

from onnxruntime.