feat(aggregate): initialize language type columns in aggregate_data function

2026-07-14 21:41:59 +00:00 · 2026-07-13 14:58:53 +02:00
parent 0be63969f3
commit 535371a5b8
1 changed files with 3 additions and 1 deletions
@@ -569,7 +569,9 @@ def aggregate_data(src_meta, dst_meta, data_idx, data_files_size_in_mb, chunk_si
            if dtype in ("video", "image"):
                continue
            n_rows = len(df)
-            if dtype == "bool":
+            if dtype == "language":
+                df[col_name] = [[] for _ in range(n_rows)]
+            elif dtype == "bool":
                df[col_name] = False
            elif dtype in ("float32", "float64"):
                df[col_name] = 0.0