Easy way to plot train and val accuracy
train loss and val loss graph.
1 Like
One simple way to plot your losses after the training would be using matplotlib:
import matplotlib.pyplot as plt
val_losses = []
train_losses = []
training loop
train_losses.append(loss_train.item())
testing
val_losses.append(loss_val.item())
plt.figure(figsize=(10,5))
plt.title("Training and Validation Loss")
plt.plot(val_losses,label="val")
plt.plot(train_losses,label="train")
plt.xlabel("iterations")
plt.ylabel("Loss")
plt.legend()
plt.show()
The more elegant way, which lets you check your progress during training as well is tensorboard:
from torch.utils.tensorboard import SummaryWriter
# Writer will output to ./runs/ directory by default
writer = SummaryWriter(logdir)
training loop
train_loss += loss_train.item()
writer.add_scalar('Loss/train', training_loss, global_step)
testing
val_loss += loss_val.item()
writer.add_scalar('Loss/val', val_loss, global_step)
writer.close()
Accuracy works the same
thank u for your reply