Index 2 is out of bounds for dimension 1 with size 2

[Applewine33]

I try to run this code while when it run to "loss = loss_function(outputs, labels)" will get error and print this RuntimeError "Index 2 is out of bounds for dimension 1 with size 2". How can I fix it?

max_epochs = 100
val_interval = 2
best_metric = -1
best_metric_epoch = -1
epoch_loss_values = []
metric_values = []
post_pred = Compose([AsDiscrete(argmax=True, to_onehot=2)])
post_label = Compose([AsDiscrete(to_onehot=2)])

for epoch in range(max_epochs):
print("-" * 10)
print(f"epoch {epoch + 1}/{max_epochs}")
model.train()
epoch_loss = 0
step = 0
for batch_data in train_loader:
step += 1
inputs, labels = (
batch_data["image"].to(device),
batch_data["label"].to(device),
)
optimizer.zero_grad()
outputs = model(inputs)
loss = loss_function(outputs, labels)
loss.backward()
optimizer.step()
epoch_loss += loss.item()
print(f"{step}/{len(train_ds) // train_loader.batch_size}, " f"train_loss: {loss.item():.4f}")
epoch_loss /= step
epoch_loss_values.append(epoch_loss)
print(f"epoch {epoch + 1} average loss: {epoch_loss:.4f}")

if (epoch + 1) % val_interval == 0:
    model.eval()
    with torch.no_grad():
        for val_data in val_loader:
            val_inputs, val_labels = (
                val_data["image"].to(device),
                val_data["label"].to(device),
            )
            roi_size = (160, 160, 160)
            sw_batch_size = 4
            val_outputs = sliding_window_inference(val_inputs, roi_size, sw_batch_size, model)
            val_outputs = [post_pred(i) for i in decollate_batch(val_outputs)]
            val_labels = [post_label(i) for i in decollate_batch(val_labels)]
            # compute metric for current iteration
            dice_metric(y_pred=val_outputs, y=val_labels)

        # aggregate the final mean dice result
        metric = dice_metric.aggregate().item()
        # reset the status for next validation round
        dice_metric.reset()

        metric_values.append(metric)
        if metric > best_metric:
            best_metric = metric
            best_metric_epoch = epoch + 1
            torch.save(model.state_dict(), os.path.join(images_dir, "best_metric_model.pth"))
            print("saved new best metric model")
        print(
            f"current epoch: {epoch + 1} current mean dice: {metric:.4f}"
            f"\nbest mean dice: {best_metric:.4f} "
            f"at epoch: {best_metric_epoch}"
        )

[KumoLiu]

Hi @Applewine33, thanks for your interest here.
The error you're encountering, RuntimeError: Index 2 is out of bounds for dimension 1 with size 2, generally happens when the code tries to access index 2 of dimension 1, but that dimension only has a size of 2 (meaning it only has valid indices 0 and 1).
In your specific case of this occurring at the line loss = loss_function(outputs, labels), this indicates that there most likely is a mismatch between the number of classes in your model output and the number of classes you have in your target labels.
Here are a few things you can check:
Output Classes: Ensure that the number of output units in the last layer of your network matches the number of classes in your dataset.
Label Values: Check your label tensor to make sure that it does not contain any values higher than (number_of_output_units - 1). If your model only outputs 2 classes (indices 0 and 1), you should check that the labels tensor does not contain any values higher than 1.

Thanks.
(more like a usage issue, move it to discussion)

[Applewine33]

I find what wrong and fix it. But while I count train loss, it shows that "The size of the proposed random crop ROI is larger than the image size". I try to change the spatial_size in RandCropByPosNegLabeld to be smaller but it do not work. How should I do?

[KumoLiu]

Ensure the ROI size smaller than the image size. Or You can set allow_smaller=True to skip the error.

MONAI/monai/transforms/croppad/dictionary.py

Line 919 in f6f9e81

allow_smaller: if `False`, an exception will be raised if the image is smaller than

[Applewine33]

I try this already. But still have the same error.