Orion model re-training by initializing model with trained model info

[lichengca]

• Orion version: 0.6.0
• Python version: 3.10
• Operating System: Windows

Description

We will run large amount of data through to train models, and will keep training with new data, do not want to do all over from very beginning. I saw the package has model save and load functions, it might not work in this case. Some models, like LSTM, come with callbacks options, may be used for retraining, but models like TadGAN and AER do not have the options.

It would be great if Orion has a feature to re-train by initializing model with previously trained model info.

Thanks,

What I Did

Paste the command(s) you ran and the output.
If there was a crash, please include the traceback here.

[sarahmish]

Hi @lichengca, thank you for your suggestion!

We definitely understand the need for this feature. I'm marking it as a feature request and it should be supported in future releases