Tensor shape problem

Hallo I got this message while training lstm model :

UserWarning: Using a target size (torch.Size([28, 7, 1, 1])) that is different to the input size (torch.Size([28, 7])). This will likely lead to incorrect results due to broadcasting. Please ensure they have the same size. I need to know how to solve this …

This is the training loop:

def train_model():
# Train the model
num_epochs = 10
for epoch in range(num_epochs):
for tra_batch in train_loader:
X_train, y_train = tra_batch
output = model(X_train)
loss = criterion(output, y_train.squeeze(-1))

    # Validate the model
    with torch.no_grad():
        val_losses = []
        for val_batch in valid_loader:
            X_val, y_val = val_batch
            val_output = model(X_val)
            val_loss = criterion(val_output, y_val.squeeze(-1))

    # Log training and validation loss with W&B (outside the validation loop)
    wandb.log({"Training Loss": loss.item(), "Validation Loss": sum(val_losses) / len(val_losses)}) 
    print(f'Epoch [{epoch}/{num_epochs}], Training Loss: {loss.item():.4f}, Validation Loss: {(sum(val_losses) / len(val_losses)):.4f}')