diff --git a/include/tvm/auto_scheduler/loop_state.h b/include/tvm/auto_scheduler/loop_state.h
index 489d058805c1f..fb9bc864f5dea 100644
--- a/include/tvm/auto_scheduler/loop_state.h
+++ b/include/tvm/auto_scheduler/loop_state.h
@@ -432,9 +432,8 @@ class State : public ObjectRef {
    * \param iter_id The iterator to be factored.
    * \param factor_iter_id The position where the new iterator is placed.
    * \param dag The original ComputeDAG of this state.
-   * \note Rfactor step will add an extra stage to the original ComputeDAG (see
-   * `te::Schedule::rfactor` for more details), a up-to-date ComputeDAG is stored in State's
-   * `current_compute_dag`.
+   * \note Rfactor step will add an extra stage to the original ComputeDAG (in the front of the
+   * target stage), a up-to-date ComputeDAG is stored in State's `current_compute_dag`.
    */
   TVM_DLL int rfactor(int stage_id, const Iterator& it, int factor_iter_id, const ComputeDAG& dag);
 
diff --git a/python/tvm/auto_scheduler/loop_state.py b/python/tvm/auto_scheduler/loop_state.py
index 291f5f0ecafa2..106cf6def87f2 100644
--- a/python/tvm/auto_scheduler/loop_state.py
+++ b/python/tvm/auto_scheduler/loop_state.py
@@ -487,8 +487,8 @@ def rfactor(self, stage, iterator, factor_iter_id):
 
         Notes
         -----
-        Rfactor step will insert an extra stage to the original ComputeDAG (see
-        `te.Schedule.rfactor` for more details).
+        Rfactor step will insert an extra stage to the original ComputeDAG (in the front of the
+        target stage).
         """
         self.state_object, new_stage_id = _ffi_api.StateRfactor(self.state_object,
                                                                 self._resolve_stage_id(stage),