diff --git a/tzrec/datasets/data_parser.py b/tzrec/datasets/data_parser.py
index 760b357..76afeac 100644
--- a/tzrec/datasets/data_parser.py
+++ b/tzrec/datasets/data_parser.py
@@ -453,14 +453,9 @@ def _to_sparse_features(
                     seq_length = length
                     key_length = input_data[f"{key}.key_lengths"]
                     # TODO: remove to_float when segment_reduce support int values
-                    try:
-                        length = torch.segment_reduce(
-                            key_length.float(), "sum", lengths=seq_length
-                        ).to(length.dtype)
-                    except Exception:
-                        import pdb
-
-                        pdb.set_trace()
+                    length = torch.segment_reduce(
+                        key_length.float(), "sum", lengths=seq_length
+                    ).to(length.dtype)
                     mulval_keys.append(key)
                     mulval_seq_lengths.append(seq_length)
                     mulval_key_lengths.append(key_length)