diff --git a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.train.py.swp b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.train.py.swp
index 382b7fa9b..e577566a4 100644
Binary files a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.train.py.swp and b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.train.py.swp differ
diff --git a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/train.py b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/train.py
index d62435804..3571434f7 100755
--- a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/train.py
+++ b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/train.py
@@ -1086,8 +1086,8 @@ def train_one_epoch(
                     f"grad_scale is too small, exiting: {cur_grad_scale}"
                 )
 
+        logging.info(loss)
         if params.batch_idx_train > 4000 and loss > 300:
-            logging.info(loss)
             wb.log({"valid/loss": 10000})
             raise RunteimError(
                     f"divergence... exiting: loss={loss}"