-
Notifications
You must be signed in to change notification settings - Fork 1
Setup an Experiment
Alexandra Puchko edited this page Apr 13, 2018
·
12 revisions
To setup an experiment, you will need to create an Experiment object.
Experiment takes four arguments:
- name (str): The name of the Experiment to be created
- load_fn (str): Name of the file in which Experiment JSON is saved
- optimizer (str): Name of the optimizer that the user would like to use. This can be set later.
- expt_id (str) : ID of already initialized experiment
The following example creates an experiment with the name data_proj using the random optimizer.
my_expmt = Experiment("data_proj", optimizer="random")
- bayesian
- random
- grid-search
set_optimizer(optimizer): Sets the optimizer for the for the experiment. Verifies that it is a valid optimizer.
save_expt(file_name): Saves this Experiment in JSON format to the specified file.
load_expt(file_name): Load the JSON in file_name into this Experiment.
set_optimizer(optimizer): Sets the optimizer the for the experiment. Verifies that it is a valid optimizer.