If you have a model ready, you can easily train it inside PerceptiLabs.
Training means to update the weights and biases of your model so that it learns to make predictions. A second phase known as validation, then checks the accuracy of the trained model on new data. PerceptiLabs trains and validates your model using the data you allocated in your Dataset Configuration Settings for training and validation.
Read on to learn how to train a model in PerceptiLabs.
Training a Model
Follow the steps below to train your model:
1) Click Run in the Modeling Tool:
2) (Optional) Adjust the training settings to customize how your model trains:
3) Click Run Model. If you've run the model before, PerceptiLabs will ask if you want to continue your training or start from scratch:
PerceptiLabs will then navigate to the Training View where you can view real-time statistics about your models training and validation performance:
Training Multiple Models in Parallel
You can optionally train other models in parallel as per the steps below:
1) Navigate to the Modeling Tool during training and select a different model (or select a different model from the Overview screen if no other model is open).
2) Click Run in the Modeling Tool for the new model you selected in Step 1. The Training View will display a new tab for the model and being training that model. You can navigate between the models during training by clicking on their tabs: