Tighten type hints for tensor arithmetic (#2550)

Summary:
Pull Request resolved: #2550

X-link: pytorch/executorch#6752

Fixes pytorch/pytorch#124015

cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy yf225 chenyang78 kadeng muchulee8 ColinPeppler amjames desertfire chauhang aakhundov

X-link: pytorch/pytorch#135392

Reviewed By: bobrenjc93, malfet

Differential Revision: D65753120

Pulled By: ezyang

fbshipit-source-id: 28ae18661b71078fa1dfdf8c60d1a75bec2f3e39

Author: fzimmermann89

torchrec/metrics/recall_session.py

@@ -205,9 +205,13 @@ def get_recall_states(
     ) -> Dict[str, torch.Tensor]:
 
         predictions_ranked = ranking_within_session(predictions, session)
+        # pyre-fixme[58]: `<` is not supported for operand types `Tensor` and
+        #  `Optional[int]`.
         predictions_labels = (predictions_ranked < self.top_threshold).to(torch.int32)
         if self.run_ranking_of_labels:
             labels_ranked = ranking_within_session(labels, session)
+            # pyre-fixme[58]: `<` is not supported for operand types `Tensor` and
+            #  `Optional[int]`.
             labels = (labels_ranked < self.top_threshold).to(torch.int32)
         num_true_pos = _calc_num_true_pos(labels, predictions_labels, weights)
         num_false_neg = _calc_num_false_neg(labels, predictions_labels, weights)