fix[chunked-prefill]: fix format

Author: wasamtc

src/parallax/server/executor/sglang_executor.py

@@ -116,7 +116,9 @@ def __init__(
         elif self.chunked_prefill_size is not None:
             self.chunked_prefill_size = chunked_prefill_size
         else:
-            self.chunked_prefill_size = max_sequence_length if max_sequence_length is not None else max_num_tokens_per_batch
+            self.chunked_prefill_size = (
+                max_sequence_length if max_sequence_length is not None else max_num_tokens_per_batch
+            )
         if self.lora_paths is not None and len(self.lora_paths) > 0:
             self.check_lora_server_args()
 

src/parallax/server/scheduler.py

@@ -355,7 +355,9 @@ def form_batch(self) -> List[Request]:
                 break
             cost = req.prompt_len
             if cost + inflight_tokens > self.max_num_tokens_per_batch:
-                logger.debug(f"prefill request {req.request_id} cost {cost} + inflight_tokens {inflight_tokens} > max_num_tokens_per_batch {self.max_num_tokens_per_batch}, breaking")
+                logger.debug(
+                    f"prefill request {req.request_id} cost {cost} + inflight_tokens {inflight_tokens} > max_num_tokens_per_batch {self.max_num_tokens_per_batch}, breaking"
+                )
                 continue
             batch.append(req)
             inflight_tokens += cost