Mutually exclusive parameters for sweeps

Can I pass exclusive parameters for a sweep? E.g. for a particular pre-trained model, I want to try learning rate values of [0.1, 0.2]. For another model I want to use [0.3, 0.4]. if I use the sweep configuration below, then grid search will try all the four learning rate values for each model. However, for model 1 - I want to use a learning rate of 0.1, 0.2 whereas, for model2, I want to use 0.3, 0.4.

project: my_project
name: grid_search
method: grid
values: [0.1, 0.2, 0.3, 0.4]
values: [‘model1’, ‘model2’]

Hi @hsingh-utd,
Are you hoping for model1 to always receive [0.1, 0.2] or would you like the sweep to pair [0.3, 0.4] with model1 at some point during the sweep?

For instance, you could say values:{(0.1, 0.2), (0.3, 0.4)] to give model1 either one of the sets at any point.

Or if you would like try different combinations of the 2 different learning rates you could use

values: [0.1, 0.2, 0.3, 0.4]
values: [0.1, 0.2, 0.3, 0.4]

Feel free to clarify your goal if I’m not pointing you in the right direction.

Thank you,

Hi @hsingh-utd, I wanted to follow up and see if you were still looking for help with this?

Thank you,

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.