[Docs] Fix warnings in mkdocs build (continued) (#24791)

Signed-off-by: Zerohertz <ohg3417@gmail.com>

vllm/distributed/eplb/eplb_state.py

@@ -337,11 +337,11 @@ class EplbState:
         Args:
             model (MixtureOfExperts): The MoE model.
             is_dummy (bool): If `True`, this is a dummy step and the load
-              metrics recorded in this forward pass will not count. Defaults
-              to `False`.
+                metrics recorded in this forward pass will not count. Defaults
+                to `False`.
             is_profile (bool): If `True`, perform a dummy rearrangement
-              with maximum communication cost. This is used in `profile_run`
-              to reserve enough memory for the communication buffer.
+                with maximum communication cost. This is used in `profile_run`
+                to reserve enough memory for the communication buffer.
             log_stats (bool): If `True`, log the expert load metrics.
 
         # Stats

vllm/distributed/eplb/rebalance_algo.py

@@ -102,14 +102,14 @@ def rebalance_experts_hierarchical(
     num_groups: int,
     num_nodes: int,
     num_gpus: int,
-):
+) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
     """
     Parameters:
         weight: [num_moe_layers, num_logical_experts]
         num_physical_experts: number of physical experts after replication
         num_groups: number of expert groups
-        num_nodes: number of server nodes, where the intra-node network
-        (e.g, NVLink) is faster
+        num_nodes: number of server nodes, where the intra-node network 
+            (e.g, NVLink) is faster
         num_gpus: number of GPUs, must be a multiple of `num_nodes`
 
     Returns: