View file src/ia/pytorch/mnist_conv_maxpool_ds.py - Download
import torch
from torch import nn
from torch.utils.data import DataLoader
from torchvision import datasets, transforms
from torchvision.transforms import ToTensor
from torch.utils.data import TensorDataset
import torch.nn.functional as F
import numpy
import sys
# sys.setrecursionlimit(2000)
# Get cpu or gpu device for training.
device = "cuda" if torch.cuda.is_available() else "cpu"
print(f"Using {device} device")
print("Loading labels ...")
# from labels import labels
# labels = torch.Tensor(labels)
# labels = labels.to(torch.long)
# train_labels = labels[0:8000]
# test_labels = labels[8000:9000]
print("Loading images ...")
# from images import images
# train_images = images[0:8000]
# test_images = images[8000:9000]
# train_images = numpy.array(train_images).astype('float32') / 255
# train_images = torch.Tensor(train_images)
# test_images = [test_images]
# test_images = numpy.array(test_images).astype('float32') / 255
# test_images = torch.Tensor(test_images)
# training_data = TensorDataset(torch.tensor(train_images,dtype=torch.float32), torch.tensor(train_labels,dtype=torch.long))
# test_data = TensorDataset(torch.tensor(test_images,dtype=torch.float32), torch.tensor(test_labels,dtype=torch.long))
transform = transforms.ToTensor()
training_data = datasets.MNIST(root='./data', train=True, download=True, transform=transform)
test_data = datasets.MNIST(root='./data', train=False, download=True, transform=transform)
batch_size = 50
# Create data loaders.
train_dataloader = DataLoader(training_data, batch_size=batch_size)
test_dataloader = DataLoader(test_data, batch_size=batch_size)
for X, y in test_dataloader:
print(f"Shape of X [N, C, H, W]: {X.shape}")
print(f"Shape of y: {y.shape} {y.dtype}")
break
# Get cpu or gpu device for training.
device = "cuda" if torch.cuda.is_available() else "cpu"
print(f"Using {device} device")
D = 28 * 28
C = 1
classes = 10
filters = 16
K = 3
# Define model
class NeuralNetwork(nn.Module):
def __init__(self):
super().__init__()
# self.flatten = nn.Flatten()
self.linear_relu_stack = nn.Sequential(
# nn.Flatten(),
# nn.Linear(28*28, 512),
# nn.ReLU(),
# nn.Linear(512, 512),
# nn.ReLU(),
# nn.Linear(512, 10)
# nn.Conv2d(C, filters, K, padding=K//2),
# nn.Tanh(),
# nn.Flatten(),
# nn.Linear(filters*D, classes),
nn.Conv2d(C, filters, K, padding=K//2),
nn.Tanh(),
nn.MaxPool2d(2),
nn.Conv2d(filters, 2*filters, K, padding=K//2),
nn.Tanh(),
nn.Flatten(),
nn.Linear(2*filters*D//4, classes),
# nn.Conv2d(C, filters, K, padding=K//2),
# nn.Tanh(),
# nn.Conv2d(filters, filters, K, padding=K//2),
# nn.Tanh(),
# nn.Conv2d(filters, filters, K, padding=K//2),
# nn.Tanh(),
# nn.MaxPool2d(2),
# nn.Conv2d(filters, 2*filters, K, padding=K//2),
# nn.Tanh(),
# nn.Conv2d(2*filters, 2*filters, K, padding=K//2),
# nn.Tanh(),
# nn.Conv2d(2*filters, 2*filters, K, padding=K//2),
# nn.Tanh(),
# nn.MaxPool2d(2),
# nn.Flatten(),
# nn.Linear(2*filters*D//(4**2), classes),
)
def forward(self, x):
# x = self.flatten(x)
logits = self.linear_relu_stack(x)
return logits
model = NeuralNetwork().to(device)
print(model)
loss_fn = nn.CrossEntropyLoss()
optimizer = torch.optim.SGD(model.parameters(), lr=1e-3)
def train(dataloader, model, loss_fn, optimizer):
size = len(dataloader.dataset)
model.train()
for batch, (X, y) in enumerate(dataloader):
X, y = X.to(device), y.to(device)
# Compute prediction error
pred = model(X)
loss = loss_fn(pred, y)
# Backpropagation
optimizer.zero_grad()
loss.backward()
optimizer.step()
if batch % 100 == 0:
loss, current = loss.item(), batch * len(X)
print(f"loss: {loss:>7f} [{current:>5d}/{size:>5d}]")
def test(dataloader, model, loss_fn):
size = len(dataloader.dataset)
num_batches = len(dataloader)
model.eval()
test_loss, correct = 0, 0
with torch.no_grad():
for X, y in dataloader:
X, y = X.to(device), y.to(device)
pred = model(X)
test_loss += loss_fn(pred, y).item()
correct += (pred.argmax(1) == y).type(torch.float32).sum().item()
test_loss /= num_batches
correct /= size
print(f"Test Error: \n Accuracy: {(100*correct):>0.1f}%, Avg loss: {test_loss:>8f} \n")
epochs = 3
for t in range(epochs):
print(f"Epoch {t+1}\n-------------------------------")
train(train_dataloader, model, loss_fn, optimizer)
test(test_dataloader, model, loss_fn)
print("Done!")
torch.save(model.state_dict(), "model.pth")
print("Saved PyTorch Model State to model.pth")
model = NeuralNetwork()
model.load_state_dict(torch.load("model.pth"))
classes = [
"Zero ",
"One ",
"Two ",
"Three",
"Four ",
"Five ",
"Six ",
"Seven",
"Eight",
"Nine ",
]
# model.eval()
model_pred = model.cpu().eval()
ngood = 0
nbad = 0
for i in range(100):
x, y = test_data[i][0], test_data[i][1]
img = x[0]
with torch.no_grad():
w, h = img.shape
x1 = img.reshape(1, -1, w, h)
logits = model(x1)
pred = F.softmax(logits, dim=1)
predicted, actual = classes[pred[0].argmax(0)], classes[y]
if predicted == actual:
result = "Good"
ngood = ngood + 1
else:
result = "Bad"
nbad = nbad + 1
print(f'Predicted: "{predicted}", Actual: "{actual}", Result: {result}')
print(f'{ngood} good, {nbad} bad')