make sharding_first by default

Author: sneaxiy

paddlenlp/trainer/training_args.py

@@ -782,7 +782,7 @@ class TrainingArguments:
                 "Following options are supported:\n"
                 "- pp_first. the topo order is dp, pp, sharding, mp \n"
                 "- sharding_first. the topo order is dp, sharding, pp, mp \n"
-                "Default is None, for pp_first"
+                "Default is None, for sharding_first"
             )
         },
     )
@@ -2072,7 +2072,7 @@ def _post_init_parallel_degree(self):
                 self.expert_tensor_parallel_degree = -1
 
         if self.hybrid_parallel_topo_order is None:
-            self.hybrid_parallel_topo_order = "pp_first"
+            self.hybrid_parallel_topo_order = "sharding_first"
         assert self.hybrid_parallel_topo_order in ["pp_first", "sharding_first"]
 
         if self.use_hybrid_parallel and self.enable_auto_parallel: