update

Author: gufengc

src/parallax/server/executor/mlx_executor.py

@@ -199,7 +199,9 @@ def __init__(
             if chunked_prefill_size == 0:
                 chunked_prefill_size = None
             elif not enable_prefix_cache:
-                raise ValueError("Chunked prefill is not supported when disable prefix cache, please enable prefix cache or set --chunked-prefill-size 0")
+                raise ValueError(
+                    "Chunked prefill is not supported when disable prefix cache, please enable prefix cache or set --chunked-prefill-size 0"
+                )
 
         self.cache_manager = CacheManager(
             num_layers=self.num_shard_layers,

src/parallax/server/shard_loader.py

@@ -39,6 +39,7 @@
     "GlmMoeDsaForCausalLM": "DeepseekV32ForCausalLM",
 }
 
+
 class MLXModelLoader:
     """
     Handles downloading model assets from Hugging Face (if needed) and loading