# train.py — REFERENCE SOLUTION (PyTorch) import torch import torch.nn as nn import numpy as np from model import MultiTaskClassifier from data import create_dataset, normalize_features, split_dataset def train_step(model, x_batch, y_batch, loss_fn, optimizer): """Single training step.""" model.train() predictions = model(x_batch) loss = loss_fn(predictions, y_batch) optimizer.zero_grad() loss.backward() optimizer.step() return loss, predictions def evaluate(model, loader, loss_fn): """Evaluate model on a DataLoader.""" model.eval() total_loss = 0.0 correct = 0 total = 0 with torch.no_grad(): for x_batch, y_batch in loader: preds = model(x_batch) loss = loss_fn(preds, y_batch) total_loss += loss.item() * len(y_batch) predicted = torch.argmax(preds, dim=1) correct += (predicted == y_batch).sum().item() total += len(y_batch) avg_loss = total_loss / total if total else 0 accuracy = correct / total if total else 0 return avg_loss, accuracy def train(model, train_loader, val_loader, epochs=10, lr=1e-3): """Full training loop.""" optimizer = torch.optim.Adam(model.parameters(), lr=lr, betas=(0.9, 0.999)) loss_fn = nn.CrossEntropyLoss() history = {"train_loss": [], "train_acc": [], "val_loss": [], "val_acc": []} for epoch in range(epochs): model.train() train_losses = [] train_correct = 0 train_total = 0 for x_batch, y_batch in train_loader: loss, preds = train_step(model, x_batch, y_batch, loss_fn, optimizer) train_losses.append(loss.item()) predicted = torch.argmax(preds, dim=1) train_correct += (predicted == y_batch).sum().item() train_total += len(y_batch) val_loss, val_acc = evaluate(model, val_loader, loss_fn) history["train_loss"].append(sum(train_losses) / len(train_losses)) history["train_acc"].append(train_correct / train_total) history["val_loss"].append(val_loss) history["val_acc"].append(val_acc) return history def save_model(model, path): """Save model weights.""" torch.save(model.state_dict(), path) if __name__ == "__main__": np.random.seed(42) torch.manual_seed(42) x = np.random.randn(200, 10).astype(np.float32) y = np.random.randint(0, 3, size=200).astype(np.int64) (x_train, y_train), (x_val, y_val) = split_dataset(x, y) x_train = normalize_features(x_train) x_val = normalize_features(x_val) train_loader = create_dataset(x_train, y_train, batch_size=16) val_loader = create_dataset(x_val, y_val, batch_size=16, shuffle=False) model = MultiTaskClassifier(input_dim=10, hidden_units=64, num_classes=3) history = train(model, train_loader, val_loader, epochs=5) save_model(model, "model.pth") print("Training complete. History:", history)