src/evaluate.py

from sklearn.metrics import classification_report, confusion_matrix
import torch

def evaluate_model(model, dataloader, device):
    model.eval()
    y_true, y_pred = [], []
    with torch.no_grad():
        for x, y in dataloader:
            x, y = x.to(device), y.to(device)
            preds = model(x).argmax(dim=1)
            y_true.extend(y.cpu().numpy())
            y_pred.extend(preds.cpu().numpy())
    return classification_report(y_true, y_pred), confusion_matrix(y_true, y_pred)