Fixed gpt-oss _load_weights_other() parameter position bug (vllm-project#28715)

Co-authored-by: Dezhan Tu <dztu@meta.com>

Author: 2 people

vllm/model_executor/models/gpt_oss.py

@@ -641,8 +641,8 @@ def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]:
             )
         else:
             return self._load_weights_other(
-                ep_rank_end,
                 ep_rank_start,
+                ep_rank_end,
                 heads_per_rank,
                 head_start,
                 weights,