2733bef356c53286d475a67476d88d4840923830,code/deep/finetune_AlexNet_ResNet/finetune_office31.py,,finetune,#,87

Before Change

    for epoch in range(1, N_EPOCH + 1):
        // lr_schedule(optimizer, epoch)
        print("Learning rate: {:.8f}".format(optimizer.param_groups[0]["lr"]))
        print("Learning rate: {:.8f}".format(optimizer.param_groups[-1]["lr"]))
        for phase in ["src", "val", "tar"]:
            if phase == "src":
                model.train()
            else:

After Change

    criterion = nn.CrossEntropyLoss()
    stop = 0
    for epoch in range(1, args.n_epoch + 1):
        stop += 1
        // You can uncomment this line for scheduling learning rate
        // lr_schedule(optimizer, epoch)
        for phase in ["src", "val", "tar"]:
            if phase == "src":
                model.train()
            else:
                model.eval()
            total_loss, correct = 0, 0
            for inputs, labels in dataloaders[phase]:
                inputs, labels = inputs.to(DEVICE), labels.to(DEVICE)
                optimizer.zero_grad()
                with torch.set_grad_enabled(phase == "src"):
                    outputs = model(inputs)
                    loss = criterion(outputs, labels)
                preds = torch.max(outputs, 1)[1]
                if phase == "src":
                    loss.backward()
                    optimizer.step()
                total_loss += loss.item() * inputs.size(0)
                correct += torch.sum(preds == labels.data)
            epoch_loss = total_loss / len(dataloaders[phase].dataset)
            epoch_acc = correct.double() / len(dataloaders[phase].dataset)
            print("Epoch: [{:02d}/{:02d}]---{}, loss: {:.6f}, acc: {:.4f}".format(epoch, args.n_epoch, phase, epoch_loss,
                                                                                  epoch_acc))
            if phase == "val" and epoch_acc > best_acc:
                stop = 0
                best_acc = epoch_acc
                torch.save(model.state_dict(), "model.pkl")
        if stop >= args.early_stop:
            break
        print()
    model.load_state_dict(torch.load("model.pkl"))
    acc_test = test(model, dataloaders["tar"])
    time_pass = time.time() - since

Instances