Saving model

[Eric-Simon-IA]

Hello,
First, thank you for your work, I've been using USAD so far but your solution seems to give similar or even better results on my data, while being easier to tune/train

I'm not really familar with models built that way, so I'm kinda stuck on one point, if you could guide me I'd be grateful:
I trained my model on colab, but need to make it available on another computer (inference on CPU only) for experimentation purpose.

Here is the struggle : I Can't find the proper way to save the model to be used elsewhere.

What I tried :

• dill/pickle/torch.save the trained model object, rearrange functions,... but there's always at least one of the dependencies that cannot be interpreted. I know it's not the best pratice to do so anyway
• Just saved the 'net' state_dict, recreate the COUTA instance and load dict but too much things are done in fit(), so I can't reload weights if the model is not fitted before

Any help will be appreciated, thank you in advance

[Eric-Simon-IA]

Well I managed to make something work now, by saving state_dict, dim and c, but not sure if it's the best way to do that

[xuhongzuo]

Thanks for raising this issue. I will add model saving and loading into the notebook file.
Yes, saving state_dict of the network, the center c, and the dimensionality dim seems to be a good solution.

[xuhongzuo]

Hi @Eric-Simon-IA, I just added model save and load to couta_algo.py. You can now assign save_model_path and load_model_path when initializing COUTA to realize saving the trained network or loading of the fitted network. I also update a section in the Readme.md.

Thanks again for raising this issue and sorry for the late update.
I will close this issue, and if you have any further problems related to this problem, please feel free to reopen it.

Cheers! Hope you have a wonderful week!

[Eric-Simon-IA]

Thank you, will look at it when I manage to get some free time.
Hope you have a nice week as well !