TensorBoard support

Before you start

• Ensure that package Tensorboard is installed.

Enable TensorBoard

1. Save your training metrics or logs using TensorFlow or PyTorch. For example:

   from torch.utils.tensorboard import SummaryWriter writer = SummaryWriter('runs/experiment_1') for i in range(10): writer.add_scalar('Loss/train', i * 0.1, i) writer.close()

   This example generates log files in the runs directory.

2. Enable TensorBoard by adding the magic command %tensorboard in a notebook cell:

   %load_ext tensorboard %tensorboard --logdir=runs