diff --git a/neps_examples/convenience/neps_tblogger_tutorial.py b/neps_examples/convenience/neps_tblogger_tutorial.py index 57a9d7f0..933c16d6 100644 --- a/neps_examples/convenience/neps_tblogger_tutorial.py +++ b/neps_examples/convenience/neps_tblogger_tutorial.py @@ -253,7 +253,7 @@ def run_pipeline_BO(lr, optim, weight_decay): "Optimizer choices are defined differently in the pipeline_space" ) - max_epochs = 9 + max_epochs = 2 # Epochs to train the model, can be parameterized as fidelity # Load the MNIST dataset for training, validation, and testing. train_loader, validation_loader, test_loader = MNIST( @@ -338,7 +338,7 @@ def run_pipeline_BO(lr, optim, weight_decay): parser.add_argument( "--max_evaluations_total", type=int, - default=10, + default=3, help="Number of different configs to train", ) args = parser.parse_args() @@ -356,13 +356,9 @@ def run_pipeline_BO(lr, optim, weight_decay): neps.run( run_pipeline=run_pipeline_BO, pipeline_space=pipeline_space_BO(), - root_directory="bayesian_optimization", - max_evaluations_total=max_evaluations_total, - searcher="bayesian_optimization", - # By default, NePS runs 10 random configurations before sampling - # from the acquisition function. We will change this behavior with - # the following keyword argument. - initial_design_size=5, + root_directory="output", + max_evaluations_total=args.max_evaluations_total, + searcher="random_search", ) """ @@ -371,7 +367,7 @@ def run_pipeline_BO(lr, optim, weight_decay): run the following command on the file created by neps root_directory. ```bash: - tensorboard --logdir bayesian_optimization + tensorboard --logdir output ``` To be able to check the visualization of tensorboard make sure to @@ -396,10 +392,10 @@ def run_pipeline_BO(lr, optim, weight_decay): the following command in your terminal: ```bash: - python neps_tblogger_tutorial.py --max_evaluations_total 15 + python neps_tblogger_tutorial.py --max_evaluations_total 10 ``` - This adds five more configurations to your search and turns off tblogger. + This adds seven more configurations to your search and turns off tblogger. By default, tblogger is on, but you can control it with `tblogger.enable()` or `tblogger.disable()` in your code." """