PyTorch 60-menit: tensor & autograd

PyTorch: Standar Emas Riset & Engineering AI

1. Tensors & Device Management

import torch

device = "cuda" if torch.cuda.is_available() else "cpu"
x = torch.randn(3, 3, requires_grad=True, device=device)

2. Automatic Differentiation (autograd)

y = x.pow(2).sum()
y.backward()
print(x.grad)  # gradien dy/dx otomatis dihitung

3. Siklus Standar Training Loop

optimizer.zero_grad()   # 1. Bersihkan gradien akumulasi
outputs = model(inputs) # 2. Forward pass
loss = criterion(outputs, labels) # 3. Hitung loss
loss.backward()         # 4. Backward pass (hitung gradien)
optimizer.step()        # 5. Update bobot