remove commented code (#1080)

Found this commented out code while debugging, this PR removes it.

Author: danielvegamyhre

torchtitan/train.py

@@ -362,7 +362,6 @@ def train_step(self, input_dict: dict[str, torch.Tensor], labels: torch.Tensor):
                 assert len(model_parts) == 1
                 pred = model_parts[0](inputs)
                 loss = self.loss_fn(pred, labels)
-                # pred.shape=(bs, seq_len, vocab_size)
                 # need to free to before bwd to avoid peaking memory
                 del pred
                 loss.backward()