Add the training and testing folders here in the same folder.

files: trainer.py config_0.py Folder name: 'Train' (Training data folder) Folder name: 'Test' (Testing data folder)

In the command line :

Dump default configurations to the config.yaml file as reference python trainer.py --print_config>config.yaml

To change the default configurations, open the config.yaml file and change the default data

nano config.yaml

Default model parameters that are configurable:

model.learning_rate: 0.001 (Can be configured in the code) model.training_dataset_folder: train (training data folder) model.testing_dataset_folder: test (testing data folder) model.batch_size: 4

To run the model with those changed hyperparameters, type : (in the command line)

python trainer.py --config config.yaml --trainer.max_epochs 200

To load the tensorboard

tensorboard --logdir ./lightning_logs

open the link that is given in the command line. Most probably it will be http://localhost:6006/.

We can visualise the training and testing loss and accuracy per step and per epoch.