Fix prediction on df with latest dask. (#6969)

2021-05-19 12:23:03 +08:00 · 2021-05-19 12:23:03 +08:00 · 7e846bb965
commit 7e846bb965
parent 6e104f0570
1 changed files with 1 additions and 1 deletions
--- a/python-package/xgboost/dask.py
+++ b/python-package/xgboost/dask.py
@ -1000,7 +1000,7 @@ async def _direct_predict_impl(  # pylint: disable=too-many-branches
    output_shape: Tuple[int, ...],
    meta: Dict[int, str],
 ) -> _DaskCollection:
-    columns = list(meta.keys())
+    columns = tuple(meta.keys())
    if len(output_shape) >= 3 and isinstance(data, dd.DataFrame):
        # Without this check, dask will finish the prediction silently even if output
        # dimension is greater than 3.  But during map_partitions, dask passes a