Logging

You can follow the progress of the training, or inspect it once training is over, using the logging capacities of PyTorch Lightning.

TODO: description of logging and checkpoints, what does the csv file looks like

CSV logger

By default, CAREamics only uses the csv logger, which outputs a CSV file with all the training metrics. It can be inspected using the CAREamist.

import matplotlib.pyplot as plt

losses = careamist.get_losses()

plt.plot(
    losses["train_epoch"], losses["train_loss"], losses["val_epoch"], losses["val_loss"]
)
plt.legend(["Train loss", "Val loss"])
plt.title("Losses")

WandB

(soon)

TensorBoard

(soon)