import torch
from torch import nn
from torch.utils.data import DataLoader, TensorDataset
device = "cuda" if torch.cuda.is_available() else "cpu"
loader = DataLoader(TensorDataset(X, y), batch_size=32, shuffle=True)
model = nn.Sequential(nn.Linear(2, 16), nn.ReLU(), nn.Linear(16, 2)).to(device)
loss_fn = nn.CrossEntropyLoss() # logits + class indices
optimizer = torch.optim.AdamW(model.parameters(), lr=1e-3)
model.train()
for xb, yb in loader:
xb, yb = xb.to(device), yb.to(device)
loss = loss_fn(model(xb), yb)
optimizer.zero_grad()
loss.backward()
optimizer.step()
model.eval()
with torch.inference_mode():
pred = model(X.to(device)).argmax(dim=1)
torch.save(model.state_dict(), "model.pt")
model.load_state_dict(torch.load("model.pt", weights_only=True))Tahák · Python · Neuronové sítě a moderní AI
PyTorch v praxi
Tenzory, autograd, moduly, DataLoader a trénovací smyčka. Přepíšeš síť z NumPy do PyTorch a poznáš chyby, které nehlásí žádná výjimka.