test check.

python-package/xgboost/sklearn.py

@@ -1967,7 +1967,7 @@ def fit(
             provide qid.
         qid :
             Query ID for each training sample.  Should have the size of n_samples.  If
-            this is set to None, then user must provide group.
+            this is set to None, then user must provide group or a special column in X.
         sample_weight :
             Query group weights
 
@@ -1988,7 +1988,8 @@ def fit(
             query groups in the ``i``-th pair in **eval_set**.
         eval_qid :
             A list in which ``eval_qid[i]`` is the array containing query ID of ``i``-th
-            pair in **eval_set**.
+            pair in **eval_set**. The special column convention in `X` applies to
+            validation datasets as well.
 
         eval_metric : str, list of str, optional
             .. deprecated:: 1.6.0
@@ -2031,15 +2032,7 @@ def fit(
                 Use `callbacks` in :py:meth:`__init__` or :py:meth:`set_params` instead.
 
         """
-        # check if group information is provided
-
         with config_context(verbosity=self.verbosity):
-            if eval_set is not None:
-                if eval_group is None and eval_qid is None:
-                    raise ValueError(
-                        "eval_group or eval_qid is required if eval_set is not None"
-                    )
-
             train_dmatrix, evals = _wrap_evaluation_matrices(
                 missing=self.missing,
                 X=X,

tests/python/test_with_sklearn.py

@@ -217,6 +217,11 @@ def test_ranking_qid_df():
     s = ranker.score(df, y)
     assert s > 0.7
 
+    # works with validation datasets as well
+    valid_df = df.copy()
+    valid_df.iloc[0, 0] = 3.0
+    ranker.fit(df, y, eval_set=[(valid_df, y)])
+
     # same as passing qid directly
     ranker = xgb.XGBRanker(n_estimators=3, eval_metric="ndcg")
     ranker.fit(X, y, qid=q)
@@ -238,6 +243,9 @@ def test_ranking_qid_df():
     results = cross_val_score(ranker, df, y)
     assert len(results) == 5
 
+    with pytest.raises(ValueError, match="Either `group` or `qid`."):
+        ranker.fit(df, y, eval_set=[(X, y)])
+
 
 def test_stacking_regression():
     from sklearn.datasets import load_diabetes