docs style

docs/modules/server.rst

@@ -1,4 +1,4 @@
-``tesseract.server & tesseract.runtime``
+``tesseract.server & runtime``
 ========================================
 
 .. automodule:: tesseract.server

tesseract/runtime/expert_backend.py

@@ -18,9 +18,9 @@ class ExpertBackend(nn.Module):
 
     :param expert: nn.Module to be wrapped into a backend. Arbitrary pytorch module with a few limitations:
 
-        * Experts must always receive the same set of \*args and \*\*kwargs and produce output tensors of same type
-        * All \*args, \*\*kwargs and outputs must be **tensors** where 0-th dimension represents to batch size
-        * We recommend using experts that are ~invariant to the order in which they process batches
+        - Experts must always receive the same set of \*args and \*\*kwargs and produce output tensors of same type
+        - All \*args, \*\*kwargs and outputs must be **tensors** where 0-th dimension represents to batch size
+        - We recommend using experts that are ~invariant to the order in which they process batches
 
     :param opt: torch optimizer to be applied on every backward call
     :param args_schema: description of positional arguments to expert.forward, list of BatchTensorProto