Fix trainer evaluation (#4363)

* fix loss calculation in evaluation

* fix evaluation on TPU when prediction_loss_only is True

src/transformers/trainer.py

@@ -679,7 +679,7 @@ class Trainer:
         model.eval()
 
         for inputs in tqdm(dataloader, desc=description):
-            has_labels = any(inputs.get(k) is not None for k in ["labels", "masked_lm_labels"])
+            has_labels = any(inputs.get(k) is not None for k in ["labels", "lm_labels", "masked_lm_labels"])
 
             for k, v in inputs.items():
                 inputs[k] = v.to(self.args.device)
@@ -703,7 +703,7 @@ class Trainer:
                     else:
                         label_ids = np.append(label_ids, inputs["labels"].detach().cpu().numpy(), axis=0)
 
-        if is_tpu_available():
+        if is_tpu_available() and preds is not None and label_ids is not None:
             # tpu-comment: Get all predictions and labels from all worker shards of eval dataset
             preds = xm.mesh_reduce("eval_preds", preds, np.concatenate)
             label_ids = xm.mesh_reduce("eval_out_label_ids", label_ids, np.concatenate)