update conf for sharding overlap in auto_parallel static (#9456)

Author: liym27

tests/test_tipc/static/auto_parallel/llama2/pretrain_config_llama2_13b/pretrain-llama2_13b.json

@@ -10,7 +10,7 @@
   "pipeline_parallel_degree": 4,
   "sharding": "stage1",
   "data_parallel_config": "enable_allreduce_avg_in_gradinent_scale gradient_sync_after_accumulate",
-  "sharding_parallel_config": "enable_stage2_overlap",
+  "sharding_parallel_config": "enable_stage1_overlap",
   "tensor_parallel_config": "enable_mp_async_allreduce",
   "pipeline_parallel_config": "enable_send_recv_overlap enable_split_backward",
   "pipeline_schedule_mode": "VPP", 

tests/test_tipc/static/auto_parallel/llama2/pretrain_config_llama2_70b/pretrain-llama2_70b.json

@@ -52,7 +52,7 @@
     "virtual_pp_degree": 5,
     "pipeline_schedule_mode": "VPP", 
     "data_parallel_config": "enable_allreduce_avg_in_gradinent_scale gradient_sync_after_accumulate",
-    "sharding_parallel_config": "split_param enable_stage1_overlap",
+    "sharding_parallel_config": "enable_stage1_overlap",
     "tensor_parallel_config": "enable_mp_async_allreduce",
     "max_seq_length": 4096,
     "to_static": true,

tests/test_tipc/static/auto_parallel/llama2/pretrain_config_llama2_7b/pretrain-llama2_7b.json

@@ -10,7 +10,7 @@
   "pipeline_parallel_degree": 1,
   "sharding": "stage1",
   "data_parallel_config": "enable_allreduce_avg_in_gradinent_scale gradient_sync_after_accumulate",
-  "sharding_parallel_config": "enable_stage2_overlap",
+  "sharding_parallel_config": "enable_stage1_overlap",
   "tensor_parallel_config": "enable_mp_async_allreduce",
   "pipeline_parallel_config": "",
   "virtual_pp_degree": 1,