TensorBoard is a way to visualize your model and various statistics during or after training.
tensorboard --logdir [logs]
--samples_per_pluginindices the number of samples to show for each tab. Non-scalar objects are sampled using reservoir sampling.
--samples_per_plugin images=10000samples approximately 10000 images.
If you're using a custom training loop (i.e. gradient tape), then you'll need to set everything up manually.
First create a
train_log_dir = os.path.join(args.checkpoint_dir, "logs", "train") train_summary_writer = tf.summary.create_file_writer(train_log_dir)
Add scalars using
with train_summary_writer.as_default(): tf.summary.scalar("training_loss", m_loss.numpy(), step=int(ckpt.step))
from torch.utils.tensorboard import SummaryWriter writer = SummaryWriter(log_dir="./runs") writer.add_scalar("train_loss", loss_np, step) # Optionally flush e.g. at checkpoints writer.flush() # Close the writer (will flush) writer.close()