some refinement on code comment (#1324)

Co-authored-by: 宋家亮_Jialiang <[email protected]>

dlrover/python/elastic_agent/torch/training.py

@@ -290,7 +290,7 @@ def _join_rendezvous(self):
     def next_rendezvous(self):
         """The handler will periodically query the world from the master until
         the world is not empty. The world is a dictionary like
-        like {0: 8, 1: 8, 2: 8} where the key is the node ID and the value is
+        {0: 8, 1: 8, 2: 8} where the key is the node ID and the value is
         the local world size. The handler can get its rank by the position
         of it node ID in the world.
         """

dlrover/trainer/torch/elastic_run.py

@@ -36,7 +36,7 @@
 auto-config will set the nnodes as the number of nodes in a job,
 nproc_per_node as the number of available GPUs. If the number of
 nodes >= 4, it will set the network-check as True. If network-check is True,
-dlrover-run will launch simple tasks on each node to check wether
+dlrover-run will launch simple tasks on each node to check whether
 the node is slow or fault.
 
 Single-node multi-worker