How to get the best params of a model after the experiment stop

[joshua-xia]

Describe the issue: I am try to use python code to read the best params of a model and then apply them to train the model. Now I hope the nni store the best params of a model in I defined file after the experiment finished or stopped so that I can read the params in it

Is there this feature?

Thanks!

Environment:

• NNI version:
• Training service (local|remote|pai|aml|etc):
• Client OS:
• Server OS (for remote mode only):
• Python version:
• PyTorch/TensorFlow version:
• Is conda/virtualenv/venv used?:
• Is running in Docker?:

Configuration:

• Experiment config (remember to remove secrets!):
• Search space:

Log message:

• nnimanager.log:
• dispatcher.log:
• nnictl stdout and stderr:

How to reproduce it?:

[joshua-xia]

I don't want search the best params from web UI or trial log manually, I wish there is a file store the best params in such as json format so that I can read it using python code and start to train my model automatically.

[MartinMashalov]

Hey there, I have solved this issue by using the export_data method. First, you run the experiment with:

experiment.run(port=8000, wait_completion=True)

Then I fetch the results with:

experiment.export_data()

I believe the export_data method returns the results in JSON. If necessary, you may have stop the experiment to retrieve the results in your terminal:

experiment.stop()

Hope that helps!

[Lijiaoa]

hi @joshua-xia Looks like this issue having a good solution to take it. Could you tell me what's your issue status? Can I close it? Expect your reply. Thanks.