Should not change your training model or data during the hyperparameters tuning!

I try to use ray-tune to conduct the hyperparameter tuning.
I have 56 cpus but there are over 1000 configurations I need to search over.
If I change the running script during grid-search, it seems those configurations have not been run will be affected.
Does anyone observe same issue? Thanks.

That mean, if you would like to tweak the model a little bit and run hyperparameter tuning, you have to wait hyperparameter tuning of your previous model finished or you have to create a new project or folder for your new model.

Could you be more specific of what is the ideal workflow you are trying to achieve and what is the problem you run into with current Ray Tune?
A script can be helpful.