mirror of
https://github.com/aladdinpersson/Machine-Learning-Collection.git
synced 2026-02-21 11:18:01 +00:00
132 lines
3.8 KiB
Python
132 lines
3.8 KiB
Python
"""
|
|
Example code of how to use mixed precision training with PyTorch. In this
|
|
case with a (very) small and simple CNN training on MNIST dataset. This
|
|
example is based on the official PyTorch documentation on mixed precision
|
|
training.
|
|
|
|
Programmed by Aladdin Persson <aladdin.persson at hotmail dot com>
|
|
* 2020-04-10 Initial programming
|
|
* 2022-12-19 Updated comments, made sure it works with latest PyTorch
|
|
|
|
"""
|
|
|
|
# Imports
|
|
import torch
|
|
import torch.nn as nn # All neural network modules, nn.Linear, nn.Conv2d, BatchNorm, Loss functions
|
|
import torch.optim as optim # For all Optimization algorithms, SGD, Adam, etc.
|
|
import torch.nn.functional as F # All functions that don't have any parameters
|
|
from torch.utils.data import (
|
|
DataLoader,
|
|
) # Gives easier dataset managment and creates mini batches
|
|
import torchvision.datasets as datasets # Has standard datasets we can import in a nice way
|
|
import torchvision.transforms as transforms # Transformations we can perform on our dataset
|
|
|
|
|
|
# Simple CNN
|
|
class CNN(nn.Module):
|
|
def __init__(self, in_channels=1, num_classes=10):
|
|
super(CNN, self).__init__()
|
|
self.conv1 = nn.Conv2d(
|
|
in_channels=1,
|
|
out_channels=420,
|
|
kernel_size=(3, 3),
|
|
stride=(1, 1),
|
|
padding=(1, 1),
|
|
)
|
|
self.pool = nn.MaxPool2d(kernel_size=(2, 2), stride=(2, 2))
|
|
self.conv2 = nn.Conv2d(
|
|
in_channels=420,
|
|
out_channels=1000,
|
|
kernel_size=(3, 3),
|
|
stride=(1, 1),
|
|
padding=(1, 1),
|
|
)
|
|
self.fc1 = nn.Linear(1000 * 7 * 7, num_classes)
|
|
|
|
def forward(self, x):
|
|
x = F.relu(self.conv1(x))
|
|
x = self.pool(x)
|
|
x = F.relu(self.conv2(x))
|
|
x = self.pool(x)
|
|
x = x.reshape(x.shape[0], -1)
|
|
x = self.fc1(x)
|
|
|
|
return x
|
|
|
|
|
|
# Set device
|
|
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
|
assert device == "cuda", "GPU not available"
|
|
|
|
# Hyperparameters
|
|
in_channel = 1
|
|
num_classes = 10
|
|
learning_rate = 3e-4
|
|
batch_size = 100
|
|
num_epochs = 5
|
|
|
|
# Load Data
|
|
train_dataset = datasets.MNIST(
|
|
root="dataset/", train=True, transform=transforms.ToTensor(), download=True
|
|
)
|
|
train_loader = DataLoader(dataset=train_dataset, batch_size=batch_size, shuffle=True)
|
|
test_dataset = datasets.MNIST(
|
|
root="dataset/", train=False, transform=transforms.ToTensor(), download=True
|
|
)
|
|
test_loader = DataLoader(dataset=test_dataset, batch_size=batch_size, shuffle=True)
|
|
|
|
# Initialize network
|
|
model = CNN().to(device)
|
|
|
|
# Loss and optimizer
|
|
criterion = nn.CrossEntropyLoss()
|
|
optimizer = optim.Adam(model.parameters(), lr=learning_rate)
|
|
|
|
# Necessary for FP16
|
|
scaler = torch.cuda.amp.GradScaler()
|
|
|
|
# Train Network
|
|
for epoch in range(num_epochs):
|
|
for batch_idx, (data, targets) in enumerate(train_loader):
|
|
# Get data to cuda if possible
|
|
data = data.to(device=device)
|
|
targets = targets.to(device=device)
|
|
|
|
# forward
|
|
with torch.cuda.amp.autocast():
|
|
scores = model(data)
|
|
loss = criterion(scores, targets)
|
|
|
|
# backward
|
|
optimizer.zero_grad()
|
|
scaler.scale(loss).backward()
|
|
scaler.step(optimizer)
|
|
scaler.update()
|
|
|
|
|
|
# Check accuracy on training & test to see how good our model
|
|
def check_accuracy(loader, model):
|
|
num_correct = 0
|
|
num_samples = 0
|
|
model.eval()
|
|
|
|
with torch.no_grad():
|
|
for x, y in loader:
|
|
x = x.to(device=device)
|
|
y = y.to(device=device)
|
|
|
|
scores = model(x)
|
|
_, predictions = scores.max(1)
|
|
num_correct += (predictions == y).sum()
|
|
num_samples += predictions.size(0)
|
|
|
|
print(
|
|
f"Got {num_correct} / {num_samples} with accuracy {float(num_correct) / float(num_samples) * 100:.2f}"
|
|
)
|
|
|
|
model.train()
|
|
|
|
|
|
check_accuracy(train_loader, model)
|
|
check_accuracy(test_loader, model)
|