beginner/introyt/trainingyt

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

Introduction || Tensors || Autograd || Building Models || TensorBoard Support || Training Models || Model Understanding

Training with PyTorch

Created On: Nov 30, 2021 | Last Updated: May 31, 2023 | Last Verified: Nov 05, 2024

Follow along with the video below or on youtube.

Introduction

In past videos, we’ve discussed and demonstrated:

• Building models with the neural network layers and functions of the torch.nn module

• The mechanics of automated gradient computation, which is central to gradient-based model training

• Using TensorBoard to visualize training progress and other activities

In this video, we’ll be adding some new tools to your inventory:

• We’ll get familiar with the dataset and dataloader abstractions, and how they ease the process of feeding data to your model during a training loop

• We’ll discuss specific loss functions and when to use them

• We’ll look at PyTorch optimizers, which implement algorithms to adjust model weights based on the outcome of a loss function

Finally, we’ll pull all of these together and see a full PyTorch training loop in action.

Dataset and DataLoader

The Dataset and DataLoader classes encapsulate the process of pulling your data from storage and exposing it to your training loop in batches.

The Dataset is responsible for accessing and processing single instances of data.

The DataLoader pulls instances of data from the Dataset (either automatically or with a sampler that you define), collects them in batches, and returns them for consumption by your training loop. The DataLoader works with all kinds of datasets, regardless of the type of data they contain.

For this tutorial, we’ll be using the Fashion-MNIST dataset provided by TorchVision. We use torchvision.transforms.Normalize() to zero-center and normalize the distribution of the image tile content, and download both training and validation data splits.

import torch
import torchvision
import torchvision.transforms as transforms

# PyTorch TensorBoard support
from torch.utils.tensorboard import SummaryWriter
from datetime import datetime


transform = transforms.Compose(
    [transforms.ToTensor(),
    transforms.Normalize((0.5,), (0.5,))])

# Create datasets for training & validation, download if necessary
training_set = torchvision.datasets.FashionMNIST('./data', train=True, transform=transform, download=True)
validation_set = torchvision.datasets.FashionMNIST('./data', train=False, transform=transform, download=True)

# Create data loaders for our datasets; shuffle for training, not for validation
training_loader = torch.utils.data.DataLoader(training_set, batch_size=4, shuffle=True)
validation_loader = torch.utils.data.DataLoader(validation_set, batch_size=4, shuffle=False)

# Class labels
classes = ('T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat',
        'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle Boot')

# Report split sizes
print('Training set has {} instances'.format(len(training_set)))
print('Validation set has {} instances'.format(len(validation_set)))

  0%|          | 0.00/26.4M [00:00<?, ?B/s]
  0%|          | 65.5k/26.4M [00:00<01:12, 364kB/s]
  1%|          | 229k/26.4M [00:00<00:38, 683kB/s]
  3%|▎         | 918k/26.4M [00:00<00:12, 2.11MB/s]
 14%|█▍        | 3.64M/26.4M [00:00<00:02, 8.02MB/s]
 26%|██▌       | 6.78M/26.4M [00:00<00:01, 12.8MB/s]
 47%|████▋     | 12.6M/26.4M [00:00<00:00, 22.2MB/s]
 60%|█████▉    | 15.8M/26.4M [00:01<00:00, 22.7MB/s]
 81%|████████  | 21.4M/26.4M [00:01<00:00, 28.4MB/s]
 94%|█████████▍| 24.9M/26.4M [00:01<00:00, 27.6MB/s]
100%|██████████| 26.4M/26.4M [00:01<00:00, 19.4MB/s]

  0%|          | 0.00/29.5k [00:00<?, ?B/s]
100%|██████████| 29.5k/29.5k [00:00<00:00, 326kB/s]

  0%|          | 0.00/4.42M [00:00<?, ?B/s]
  1%|▏         | 65.5k/4.42M [00:00<00:12, 358kB/s]
  5%|▌         | 229k/4.42M [00:00<00:06, 674kB/s]
 21%|██        | 918k/4.42M [00:00<00:01, 2.08MB/s]
 83%|████████▎ | 3.67M/4.42M [00:00<00:00, 7.19MB/s]
100%|██████████| 4.42M/4.42M [00:00<00:00, 6.03MB/s]

  0%|          | 0.00/5.15k [00:00<?, ?B/s]
100%|██████████| 5.15k/5.15k [00:00<00:00, 47.6MB/s]
Training set has 60000 instances
Validation set has 10000 instances

As always, let’s visualize the data as a sanity check:

import matplotlib.pyplot as plt
import numpy as np

# Helper function for inline image display
def matplotlib_imshow(img, one_channel=False):
    if one_channel:
        img = img.mean(dim=0)
    img = img / 2 + 0.5     # unnormalize
    npimg = img.numpy()
    if one_channel:
        plt.imshow(npimg, cmap="Greys")
    else:
        plt.imshow(np.transpose(npimg, (1, 2, 0)))

dataiter = iter(training_loader)
images, labels = next(dataiter)

# Create a grid from the images and show them
img_grid = torchvision.utils.make_grid(images)
matplotlib_imshow(img_grid, one_channel=True)
print('  '.join(classes[labels[j]] for j in range(4)))

Pullover  Dress  Dress  Ankle Boot

The Model

The model we’ll use in this example is a variant of LeNet-5 - it should be familiar if you’ve watched the previous videos in this series.

import torch.nn as nn
import torch.nn.functional as F

# PyTorch models inherit from torch.nn.Module
class GarmentClassifier(nn.Module):
    def __init__(self):
        super(GarmentClassifier, self).__init__()
        self.conv1 = nn.Conv2d(1, 6, 5)
        self.pool = nn.MaxPool2d(2, 2)
        self.conv2 = nn.Conv2d(6, 16, 5)
        self.fc1 = nn.Linear(16 * 4 * 4, 120)
        self.fc2 = nn.Linear(120, 84)
        self.fc3 = nn.Linear(84, 10)

    def forward(self, x):
        x = self.pool(F.relu(self.conv1(x)))
        x = self.pool(F.relu(self.conv2(x)))
        x = x.view(-1, 16 * 4 * 4)
        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)
        return x


model = GarmentClassifier()

Loss Function

For this example, we’ll be using a cross-entropy loss. For demonstration purposes, we’ll create batches of dummy output and label values, run them through the loss function, and examine the result.

loss_fn = torch.nn.CrossEntropyLoss()

# NB: Loss functions expect data in batches, so we're creating batches of 4
# Represents the model's confidence in each of the 10 classes for a given input
dummy_outputs = torch.rand(4, 10)
# Represents the correct class among the 10 being tested
dummy_labels = torch.tensor([1, 5, 3, 7])

print(dummy_outputs)
print(dummy_labels)

loss = loss_fn(dummy_outputs, dummy_labels)
print('Total loss for this batch: {}'.format(loss.item()))

tensor([[0.5725, 0.0701, 0.7984, 0.7248, 0.6974, 0.7434, 0.6711, 0.4446, 0.0455,
         0.7280],
        [0.8428, 0.2333, 0.2198, 0.9558, 0.8582, 0.3675, 0.5048, 0.7045, 0.5341,
         0.2858],
        [0.7080, 0.1596, 0.6911, 0.0654, 0.7256, 0.5347, 0.0221, 0.6075, 0.6316,
         0.1355],
        [0.9750, 0.6543, 0.0877, 0.7354, 0.2457, 0.5542, 0.5423, 0.2040, 0.1801,
         0.0383]])
tensor([1, 5, 3, 7])
Total loss for this batch: 2.6504311561584473

Optimizer

For this example, we’ll be using simple stochastic gradient descent with momentum.

It can be instructive to try some variations on this optimization scheme:

• Learning rate determines the size of the steps the optimizer takes. What does a different learning rate do to the your training results, in terms of accuracy and convergence time?

• Momentum nudges the optimizer in the direction of strongest gradient over multiple steps. What does changing this value do to your results?

• Try some different optimization algorithms, such as averaged SGD, Adagrad, or Adam. How do your results differ?

# Optimizers specified in the torch.optim package
optimizer = torch.optim.SGD(model.parameters(), lr=0.001, momentum=0.9)

The Training Loop

Below, we have a function that performs one training epoch. It enumerates data from the DataLoader, and on each pass of the loop does the following:

• Gets a batch of training data from the DataLoader

• Zeros the optimizer’s gradients

• Performs an inference - that is, gets predictions from the model for an input batch

• Calculates the loss for that set of predictions vs. the labels on the dataset

• Calculates the backward gradients over the learning weights

• Tells the optimizer to perform one learning step - that is, adjust the model’s learning weights based on the observed gradients for this batch, according to the optimization algorithm we chose

• It reports on the loss for every 1000 batches.

• Finally, it reports the average per-batch loss for the last 1000 batches, for comparison with a validation run

def train_one_epoch(epoch_index, tb_writer):
    running_loss = 0.
    last_loss = 0.

    # Here, we use enumerate(training_loader) instead of
    # iter(training_loader) so that we can track the batch
    # index and do some intra-epoch reporting
    for i, data in enumerate(training_loader):
        # Every data instance is an input + label pair
        inputs, labels = data

        # Zero your gradients for every batch!
        optimizer.zero_grad()

        # Make predictions for this batch
        outputs = model(inputs)

        # Compute the loss and its gradients
        loss = loss_fn(outputs, labels)
        loss.backward()

        # Adjust learning weights
        optimizer.step()

        # Gather data and report
        running_loss += loss.item()
        if i % 1000 == 999:
            last_loss = running_loss / 1000 # loss per batch
            print('  batch {} loss: {}'.format(i + 1, last_loss))
            tb_x = epoch_index * len(training_loader) + i + 1
            tb_writer.add_scalar('Loss/train', last_loss, tb_x)
            running_loss = 0.

    return last_loss

Per-Epoch Activity

There are a couple of things we’ll want to do once per epoch:

• Perform validation by checking our relative loss on a set of data that was not used for training, and report this

• Save a copy of the model

Here, we’ll do our reporting in TensorBoard. This will require going to the command line to start TensorBoard, and opening it in another browser tab.

# Initializing in a separate cell so we can easily add more epochs to the same run
timestamp = datetime.now().strftime('%Y%m%d_%H%M%S')
writer = SummaryWriter('runs/fashion_trainer_{}'.format(timestamp))
epoch_number = 0

EPOCHS = 5

best_vloss = 1_000_000.

for epoch in range(EPOCHS):
    print('EPOCH {}:'.format(epoch_number + 1))

    # Make sure gradient tracking is on, and do a pass over the data
    model.train(True)
    avg_loss = train_one_epoch(epoch_number, writer)


    running_vloss = 0.0
    # Set the model to evaluation mode, disabling dropout and using population
    # statistics for batch normalization.
    model.eval()

    # Disable gradient computation and reduce memory consumption.
    with torch.no_grad():
        for i, vdata in enumerate(validation_loader):
            vinputs, vlabels = vdata
            voutputs = model(vinputs)
            vloss = loss_fn(voutputs, vlabels)
            running_vloss += vloss

    avg_vloss = running_vloss / (i + 1)
    print('LOSS train {} valid {}'.format(avg_loss, avg_vloss))

    # Log the running loss averaged per batch
    # for both training and validation
    writer.add_scalars('Training vs. Validation Loss',
                    { 'Training' : avg_loss, 'Validation' : avg_vloss },
                    epoch_number + 1)
    writer.flush()

    # Track best performance, and save the model's state
    if avg_vloss < best_vloss:
        best_vloss = avg_vloss
        model_path = 'model_{}_{}'.format(timestamp, epoch_number)
        torch.save(model.state_dict(), model_path)

    epoch_number += 1

EPOCH 1:
  batch 1000 loss: 2.2521325249671937
  batch 2000 loss: 0.9854257740704343
  batch 3000 loss: 0.6851327550658025
  batch 4000 loss: 0.6135343031922821
  batch 5000 loss: 0.5866948459818959
  batch 6000 loss: 0.5511749164620414
  batch 7000 loss: 0.5276467088288628
  batch 8000 loss: 0.4859999405753333
  batch 9000 loss: 0.47852722579229157
  batch 10000 loss: 0.4600267232111655
  batch 11000 loss: 0.45475304189126475
  batch 12000 loss: 0.4281130772959441
  batch 13000 loss: 0.42768952836247626
  batch 14000 loss: 0.41485296625754564
  batch 15000 loss: 0.3940568761495815
LOSS train 0.3940568761495815 valid 0.3956470787525177
EPOCH 2:
  batch 1000 loss: 0.38715169560920915
  batch 2000 loss: 0.3838005330477026
  batch 3000 loss: 0.3631052174879587
  batch 4000 loss: 0.37617891634977424
  batch 5000 loss: 0.39103491323449996
  batch 6000 loss: 0.36566069163393694
  batch 7000 loss: 0.3667216809855308
  batch 8000 loss: 0.35283140427662874
  batch 9000 loss: 0.3527607740295061
  batch 10000 loss: 0.34925047580266255
  batch 11000 loss: 0.37366600868930255
  batch 12000 loss: 0.34449519611988216
  batch 13000 loss: 0.3459661583202251
  batch 14000 loss: 0.33271378462779105
  batch 15000 loss: 0.3584383065847287
LOSS train 0.3584383065847287 valid 0.3456021845340729
EPOCH 3:
  batch 1000 loss: 0.3251506279369933
  batch 2000 loss: 0.3095294443309249
  batch 3000 loss: 0.34288687394879525
  batch 4000 loss: 0.32768870879523454
  batch 5000 loss: 0.32413441178604263
  batch 6000 loss: 0.32304036018396437
  batch 7000 loss: 0.3266114277824272
  batch 8000 loss: 0.3210621049029942
  batch 9000 loss: 0.3104973297845281
  batch 10000 loss: 0.30057946301011546
  batch 11000 loss: 0.31575276248247247
  batch 12000 loss: 0.3107170149516714
  batch 13000 loss: 0.30591063810196645
  batch 14000 loss: 0.3338698188686976
  batch 15000 loss: 0.3056864722693572
LOSS train 0.3056864722693572 valid 0.33726003766059875
EPOCH 4:
  batch 1000 loss: 0.3011234230146365
  batch 2000 loss: 0.2842651428124882
  batch 3000 loss: 0.29927633204415727
  batch 4000 loss: 0.30857997825050554
  batch 5000 loss: 0.2945819073669918
  batch 6000 loss: 0.29021309380311866
  batch 7000 loss: 0.2970107369068719
  batch 8000 loss: 0.306313495246055
  batch 9000 loss: 0.30915809172308945
  batch 10000 loss: 0.27482260235446304
  batch 11000 loss: 0.2918123505726398
  batch 12000 loss: 0.30168602007618756
  batch 13000 loss: 0.28233931581465005
  batch 14000 loss: 0.2691737981650222
  batch 15000 loss: 0.2885731645249907
LOSS train 0.2885731645249907 valid 0.31354424357414246
EPOCH 5:
  batch 1000 loss: 0.272097287611592
  batch 2000 loss: 0.2746666179396416
  batch 3000 loss: 0.2689908763434651
  batch 4000 loss: 0.26544726403536467
  batch 5000 loss: 0.27658389077588436
  batch 6000 loss: 0.27193892943436365
  batch 7000 loss: 0.26244054821689494
  batch 8000 loss: 0.30141668145651784
  batch 9000 loss: 0.28003143241122236
  batch 10000 loss: 0.2728267575435306
  batch 11000 loss: 0.264361325072352
  batch 12000 loss: 0.2864540826071679
  batch 13000 loss: 0.2513803336042256
  batch 14000 loss: 0.28345379100979246
  batch 15000 loss: 0.26518901752027796
LOSS train 0.26518901752027796 valid 0.3259730935096741

To load a saved version of the model:

saved_model = GarmentClassifier()
saved_model.load_state_dict(torch.load(PATH))

Once you’ve loaded the model, it’s ready for whatever you need it for - more training, inference, or analysis.

Note that if your model has constructor parameters that affect model structure, you’ll need to provide them and configure the model identically to the state in which it was saved.

Other Resources

• Docs on the data utilities, including Dataset and DataLoader, at pytorch.org

• A note on the use of pinned memory for GPU training

• Documentation on the datasets available in TorchVision, TorchText, and TorchAudio

• Documentation on the loss functions available in PyTorch

• Documentation on the torch.optim package, which includes optimizers and related tools, such as learning rate scheduling

• A detailed tutorial on saving and loading models

• The Tutorials section of pytorch.org contains tutorials on a broad variety of training tasks, including classification in different domains, generative adversarial networks, reinforcement learning, and more