fix(tsv): skip empty lines

mvl-sib200.py

@@ -8,6 +8,8 @@ import datasets
 import numpy as np
 import pandas as pd
 
+file_path = "/network/scratch/s/schmidtf/.cache/huggingface/hub/datasets--wuenlp--mvl-sib200/snapshots/0b12f60218de86836d2f048f12b13b8eb7ab7ac2/data/sib200/arb_Latn/train.tsv"
+
 # fmt: off
 LANGS = [
     "ace_Arab", "ace_Latn", "acm_Arab", "acq_Arab", "aeb_Arab", "afr_Latn", "ajp_Arab", 
@@ -136,6 +138,12 @@ def read_tsv_to_dict_list(file_path: Union[str, Path]) -> List[Dict[str, Any]]:
 
         # Start enumerating from line 2 to account for the header line
         for _, row in enumerate(reader, start=2):
+            #
+            if all(
+                (row[key].strip() == key) or (row[key].strip() == "")
+                for key in expected_headers
+            ):
+                continue
             # Convert index_id to integer
             index_id = int(row["index_id"])
             # Strip leading/trailing whitespace