[Docs] Fix warnings in vllm/profiler and vllm/transformers_utils (#25220)

Signed-off-by: windsonsea <haifeng.yao@daocloud.io>

mkdocs.yaml

@@ -102,6 +102,7 @@ plugins:
           - https://numpy.org/doc/stable/objects.inv
           - https://pytorch.org/docs/stable/objects.inv
           - https://psutil.readthedocs.io/en/stable/objects.inv
+          - https://huggingface.co/docs/transformers/main/en/objects.inv
 
 markdown_extensions:
   - attr_list

vllm/profiler/layerwise_profile.py

@@ -353,8 +353,8 @@ class layerwise_profile(profile):
 
         Args:
             num_running_seqs (Optional[int], optional): When given,
-            num_running_seqs will be passed to LayerProfileResults for metadata
-            update. Defaults to None.
+                num_running_seqs will be passed to LayerProfileResults
+                for metadata update. Defaults to None.
         """
         super().__init__(
             activities=[ProfilerActivity.CPU, ProfilerActivity.CUDA],

vllm/transformers_utils/configs/jais.py

@@ -74,8 +74,7 @@ class JAISConfig(PretrainedConfig):
         use_cache (`bool`, *optional*, defaults to `True`):
             Whether or not the model should return the last key/values
             attentions (not used by all models).
-        scale_attn_by_inverse_layer_idx
-            (`bool`, *optional*, defaults to `False`): 
+        scale_attn_by_inverse_layer_idx (`bool`, *optional*, default `True`):
             Whether to additionally scale attention weights 
             by `1 / layer_idx + 1`.
         reorder_and_upcast_attn (`bool`, *optional*, defaults to `False`):