Shortcuts

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

Training with PyTorch

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)))
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-images-idx3-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-images-idx3-ubyte.gz to ./data/FashionMNIST/raw/train-images-idx3-ubyte.gz

  0%|          | 0/26421880 [00:00<?, ?it/s]
  0%|          | 32768/26421880 [00:00<01:25, 308679.55it/s]
  0%|          | 65536/26421880 [00:00<01:26, 303382.44it/s]
  0%|          | 131072/26421880 [00:00<00:59, 439721.58it/s]
  1%|          | 229376/26421880 [00:00<00:42, 623022.00it/s]
  2%|1         | 491520/26421880 [00:00<00:20, 1265542.10it/s]
  4%|3         | 950272/26421880 [00:00<00:11, 2266368.00it/s]
  7%|7         | 1933312/26421880 [00:00<00:05, 4470167.18it/s]
 15%|#4        | 3833856/26421880 [00:00<00:02, 8596671.05it/s]
 26%|##6       | 6946816/26421880 [00:00<00:01, 14824670.23it/s]
 38%|###8      | 10092544/26421880 [00:01<00:00, 19078892.85it/s]
 50%|####9     | 13172736/26421880 [00:01<00:00, 21942920.58it/s]
 62%|######1   | 16285696/26421880 [00:01<00:00, 23935780.66it/s]
 73%|#######3  | 19398656/26421880 [00:01<00:00, 25077724.15it/s]
 85%|########5 | 22478848/26421880 [00:01<00:00, 26247960.23it/s]
 97%|#########6| 25591808/26421880 [00:01<00:00, 26913657.01it/s]
100%|##########| 26421880/26421880 [00:01<00:00, 16102401.52it/s]
Extracting ./data/FashionMNIST/raw/train-images-idx3-ubyte.gz to ./data/FashionMNIST/raw

Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-labels-idx1-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-labels-idx1-ubyte.gz to ./data/FashionMNIST/raw/train-labels-idx1-ubyte.gz

  0%|          | 0/29515 [00:00<?, ?it/s]
100%|##########| 29515/29515 [00:00<00:00, 267401.55it/s]
100%|##########| 29515/29515 [00:00<00:00, 266176.90it/s]
Extracting ./data/FashionMNIST/raw/train-labels-idx1-ubyte.gz to ./data/FashionMNIST/raw

Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz to ./data/FashionMNIST/raw/t10k-images-idx3-ubyte.gz

  0%|          | 0/4422102 [00:00<?, ?it/s]
  1%|          | 32768/4422102 [00:00<00:14, 302596.34it/s]
  1%|1         | 65536/4422102 [00:00<00:14, 301113.27it/s]
  3%|2         | 131072/4422102 [00:00<00:09, 438037.38it/s]
  5%|5         | 229376/4422102 [00:00<00:06, 621395.57it/s]
 11%|#1        | 491520/4422102 [00:00<00:03, 1264045.09it/s]
 21%|##1       | 950272/4422102 [00:00<00:01, 2263747.29it/s]
 44%|####3     | 1933312/4422102 [00:00<00:00, 4469942.85it/s]
 87%|########6 | 3833856/4422102 [00:00<00:00, 8596257.22it/s]
100%|##########| 4422102/4422102 [00:00<00:00, 5056649.89it/s]
Extracting ./data/FashionMNIST/raw/t10k-images-idx3-ubyte.gz to ./data/FashionMNIST/raw

Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-labels-idx1-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-labels-idx1-ubyte.gz to ./data/FashionMNIST/raw/t10k-labels-idx1-ubyte.gz

  0%|          | 0/5148 [00:00<?, ?it/s]
100%|##########| 5148/5148 [00:00<00:00, 26077629.22it/s]
Extracting ./data/FashionMNIST/raw/t10k-labels-idx1-ubyte.gz to ./data/FashionMNIST/raw

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)))
trainingyt
T-shirt/top  Pullover  Pullover  Shirt

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([[0.0605, 0.7764, 0.0904, 0.8244, 0.9021, 0.6213, 0.3305, 0.8705, 0.6494,
         0.9675],
        [0.7482, 0.9490, 0.1312, 0.0553, 0.7820, 0.9068, 0.0701, 0.9655, 0.6281,
         0.2322],
        [0.2391, 0.5261, 0.9372, 0.1935, 0.8822, 0.7540, 0.8811, 0.4726, 0.0354,
         0.8377],
        [0.1719, 0.2086, 0.6185, 0.9612, 0.8309, 0.7191, 0.3401, 0.7682, 0.7015,
         0.9431]])
tensor([1, 5, 3, 7])
Total loss for this batch: 2.278627395629883

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)

    # We don't need gradients on to do reporting
    model.train(False)

    running_vloss = 0.0
    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: 1.935198988199234
  batch 2000 loss: 0.8305015639662743
  batch 3000 loss: 0.6696620054626837
  batch 4000 loss: 0.6164027689406648
  batch 5000 loss: 0.5512013011863455
  batch 6000 loss: 0.5094508152359339
  batch 7000 loss: 0.5010407641686033
  batch 8000 loss: 0.49061456594918856
  batch 9000 loss: 0.47273052378138525
  batch 10000 loss: 0.4672327147938777
  batch 11000 loss: 0.4411310525578447
  batch 12000 loss: 0.42955538784293457
  batch 13000 loss: 0.4097148931053816
  batch 14000 loss: 0.42148528807249386
  batch 15000 loss: 0.41105559648867346
LOSS train 0.41105559648867346 valid 0.4415995180606842
EPOCH 2:
  batch 1000 loss: 0.4007492928595457
  batch 2000 loss: 0.3906503771022544
  batch 3000 loss: 0.38429762851832494
  batch 4000 loss: 0.37758284392708447
  batch 5000 loss: 0.3733714599085506
  batch 6000 loss: 0.3427473809006042
  batch 7000 loss: 0.35961880366434346
  batch 8000 loss: 0.35050802313326856
  batch 9000 loss: 0.3739806862882106
  batch 10000 loss: 0.3452117069855449
  batch 11000 loss: 0.35297094938752704
  batch 12000 loss: 0.3602626284995931
  batch 13000 loss: 0.3426916321496246
  batch 14000 loss: 0.3482986818345089
  batch 15000 loss: 0.3239242781752837
LOSS train 0.3239242781752837 valid 0.3678557276725769
EPOCH 3:
  batch 1000 loss: 0.3140595094627497
  batch 2000 loss: 0.32369600109680324
  batch 3000 loss: 0.3225217724496615
  batch 4000 loss: 0.3173451621191343
  batch 5000 loss: 0.3119620620815258
  batch 6000 loss: 0.3079630302180885
  batch 7000 loss: 0.300567087161704
  batch 8000 loss: 0.31275426331895867
  batch 9000 loss: 0.3064340375015745
  batch 10000 loss: 0.30798760230492916
  batch 11000 loss: 0.31648085953161353
  batch 12000 loss: 0.3371193317830621
  batch 13000 loss: 0.31529894701269223
  batch 14000 loss: 0.307852462017996
  batch 15000 loss: 0.31205490061937596
LOSS train 0.31205490061937596 valid 0.3237225115299225
EPOCH 4:
  batch 1000 loss: 0.28421558979798284
  batch 2000 loss: 0.2832156135141704
  batch 3000 loss: 0.29388247846056764
  batch 4000 loss: 0.31038088884713944
  batch 5000 loss: 0.27907642129203303
  batch 6000 loss: 0.29532576539077854
  batch 7000 loss: 0.2868270348364922
  batch 8000 loss: 0.29239611503811463
  batch 9000 loss: 0.2906981737676542
  batch 10000 loss: 0.2848005289405119
  batch 11000 loss: 0.285830237467766
  batch 12000 loss: 0.29427837575129523
  batch 13000 loss: 0.28754894772462286
  batch 14000 loss: 0.28905848667302053
  batch 15000 loss: 0.285255809801165
LOSS train 0.285255809801165 valid 0.29989075660705566
EPOCH 5:
  batch 1000 loss: 0.26493522461219254
  batch 2000 loss: 0.2561846436517517
  batch 3000 loss: 0.27185415715797717
  batch 4000 loss: 0.27086156053382726
  batch 5000 loss: 0.27270630368590354
  batch 6000 loss: 0.28139493602483706
  batch 7000 loss: 0.2596448304573814
  batch 8000 loss: 0.2712655549228948
  batch 9000 loss: 0.2651794503508863
  batch 10000 loss: 0.27985832289266543
  batch 11000 loss: 0.27843118878346285
  batch 12000 loss: 0.2707799100063403
  batch 13000 loss: 0.2693156594813481
  batch 14000 loss: 0.26688308830060126
  batch 15000 loss: 0.26343625859207986
LOSS train 0.26343625859207986 valid 0.3301874101161957

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

Total running time of the script: ( 4 minutes 26.715 seconds)

Gallery generated by Sphinx-Gallery

Docs

Access comprehensive developer documentation for PyTorch

View Docs

Tutorials

Get in-depth tutorials for beginners and advanced developers

View Tutorials

Resources

Find development resources and get your questions answered

View Resources