Chapter 33
09. PyTorch Model Deployment Exercises
NotebookPython 3.8.1328 cells
09. PyTorch Model Deployment Exercises
Welcome to the 09. PyTorch Model Deployment exercises.
Your objective is to write code to satisify each of the exercises below.
Some starter code has been provided to make sure you have all the resources you need.
Note: There may be more than one solution to each of the exercises.
Resources
- These exercises/solutions are based on section 09. PyTorch Model Deployment of the Learn PyTorch for Deep Learning course by Zero to Mastery.
- See a live walkthrough of the solutions (errors and all) on YouTube (but try the exercises yourself first!).
- See all solutions on the course GitHub.
Note: The first section of this notebook is dedicated to getting various helper functions and datasets used for the exercises. The exercises start at the heading "Exercise 1: ...".
Get various imports and helper functions
The code in the following cells prepares imports and data for the exercises below. They are taken from 09. PyTorch Model Deployment.
In [1]python · cell 4
python
# For this notebook to run with updated APIs, we need torch 1.12+ and torchvision 0.13+
try:
import torch
import torchvision
assert int(torch.__version__.split(".")[1]) >= 12 or int(torch.__version__.split(".")[0]) > 1, "torch version should be 1.12+"
assert int(torchvision.__version__.split(".")[1]) >= 13, "torchvision version should be 0.13+"
print(f"torch version: {torch.__version__}")
print(f"torchvision version: {torchvision.__version__}")
except:
print(f"[INFO] torch/torchvision versions not as required, installing nightly versions.")
!pip3 install -U torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu113
import torch
import torchvision
print(f"torch version: {torch.__version__}")
print(f"torchvision version: {torchvision.__version__}")Output
torch version: 1.12.1+cu113 torchvision version: 0.13.1+cu113
In [2]python · cell 5
python
# Continue with regular imports
import matplotlib.pyplot as plt
import torch
import torchvision
from torch import nn
from torchvision import transforms
# Try to get torchinfo, install it if it doesn't work
try:
from torchinfo import summary
except:
print("[INFO] Couldn't find torchinfo... installing it.")
!pip install -q torchinfo
from torchinfo import summary
# Try to import the going_modular directory, download it from GitHub if it doesn't work
try:
from going_modular.going_modular import data_setup, engine
from helper_functions import download_data, set_seeds, plot_loss_curves
except:
# Get the going_modular scripts
print("[INFO] Couldn't find going_modular or helper_functions scripts... downloading them from GitHub.")
!git clone https://github.com/mrdbourke/pytorch-deep-learning
!mv pytorch-deep-learning/going_modular .
!mv pytorch-deep-learning/helper_functions.py . # get the helper_functions.py script
!rm -rf pytorch-deep-learning
from going_modular.going_modular import data_setup, engine
from helper_functions import download_data, set_seeds, plot_loss_curvesIn [3]python · cell 6
python
device = "cuda" if torch.cuda.is_available() else "cpu"
deviceOutput
'cuda'
Get data
Want to download the data we've been using in PyTorch Model Deployment: https://www.learnpytorch.io/09_pytorch_model_deployment/#1-getting-data
In [4]python · cell 8
python
# Download pizza, steak, sushi images from GitHub
image_path = download_data(source="https://github.com/mrdbourke/pytorch-deep-learning/raw/main/data/pizza_steak_sushi_20_percent.zip",
destination="pizza_steak_sushi")
image_pathOutput
[INFO] data/pizza_steak_sushi directory exists, skipping download.
PosixPath('data/pizza_steak_sushi')In [5]python · cell 9
python
# Setup directory paths to train and test images
train_dir = image_path / "train"
test_dir = image_path / "test"Preprocess data
Turn images into tensors using same code as PyTorch Paper Replicating section 2.1 and 2.2: https://www.learnpytorch.io/08_pytorch_paper_replicating/#21-prepare-transforms-for-images
In [6]python · cell 11
python
# Create image size (from Table 3 in the ViT paper)
IMG_SIZE = 224
# Create transform pipeline manually
manual_transforms = transforms.Compose([
transforms.Resize((IMG_SIZE, IMG_SIZE)),
transforms.ToTensor(),
])
print(f"Manually created transforms: {manual_transforms}")Output
Manually created transforms: Compose(
Resize(size=(224, 224), interpolation=bilinear, max_size=None, antialias=None)
ToTensor()
)
In [7]python · cell 12
python
# Set the batch size
BATCH_SIZE = 32 # this is lower than the ViT paper but it's because we're starting small
# Create data loaders
train_dataloader, test_dataloader, class_names = data_setup.create_dataloaders(
train_dir=train_dir,
test_dir=test_dir,
transform=manual_transforms, # use manually created transforms
batch_size=BATCH_SIZE
)
train_dataloader, test_dataloader, class_namesOutput
(<torch.utils.data.dataloader.DataLoader at 0x7fb52d4b77d0>, <torch.utils.data.dataloader.DataLoader at 0x7fb52d46a450>, ['pizza', 'steak', 'sushi'])
In [8]python · cell 13
python
# Get a batch of images
image_batch, label_batch = next(iter(train_dataloader))
# Get a single image from the batch
image, label = image_batch[0], label_batch[0]
# View the batch shapes
image.shape, labelOutput
(torch.Size([3, 224, 224]), tensor(2))
In [9]python · cell 14
python
# Plot image with matplotlib
plt.imshow(image.permute(1, 2, 0)) # rearrange image dimensions to suit matplotlib [color_channels, height, width] -> [height, width, color_channels]
plt.title(class_names[label])
plt.axis(False);Output
<Figure size 432x288 with 1 Axes>
[省略较大 image/png 输出]
Exercise 1. Make and time predictions with both feature extractor models on the test dataset using the GPU (device="cuda").
- Compare the model's prediction times on GPU vs CPU - does this close the gap between them? As in, does making predictions on the GPU make the ViT feature extractor prediction times closer to the EffNetB2 feature extractor prediction times?
- You'll find code to do these steps in section 5. Making predictions with our trained models and timing them and section 6. Comparing model results, prediction times and size.
In [10]python · cell 16
python
# TODO: your codeExercise 2. The ViT feature extractor seems to have more learning capacity (due to more parameters) than EffNetB2, how does it go on the larger 20% split of the entire Food101 dataset?
- Train a ViT feature extractor on the 20% Food101 dataset for 5 epochs, just like we did with EffNetB2 in section 10. Creating FoodVision Big.
In [11]python · cell 18
python
# TODO: your codeExercise 3. Make predictions across the 20% Food101 test dataset with the ViT feature extractor from exercise 2 and find the "most wrong" predictions
- The predictions will be the ones with the highest prediction probability but with the wrong predicted label.
- Write a sentence or two about why you think the model got these predictions wrong.
In [12]python · cell 20
python
# TODO: your codeExercise 4. Evaluate the ViT feature extractor across the whole Food101 test dataset rather than just the 20% version, how does it perform?
- Does it beat the original Food101 paper's best result of 56.4% accuracy?
In [13]python · cell 22
python
# TODO: your codeExercise 5. Head to Paperswithcode.com and find the current best performing model on the Food101 dataset.
- What model architecture does it use?
In [14]python · cell 24
python
# TODO: your answer to the aboveExercise 6. Write down 1-3 potential failure points of our deployed FoodVision models and what some potential solutions might be.
- For example, what happens if someone was to upload a photo that wasn't of food to our FoodVision Mini model?
In [15]python · cell 26
python
# TODO: your answer to the aboveExercise 7. Pick any dataset from torchvision.datasets and train a feature extractor model on it using a model from torchvision.models (you could use one of the model's we've already created, e.g. EffNetB2 or ViT) for 5 epochs and then deploy your model as a Gradio app to Hugging Face Spaces.
- You may want to pick smaller dataset/make a smaller split of it so training doesn't take too long.
- I'd love to see your deployed models! So be sure to share them in Discord or on the course GitHub Discussions page.
In [16]python · cell 28
python
# TODO: your code