fix backend

vllm_plugin/scripts/start_server.py

@@ -90,9 +90,9 @@ def start_vllm_server(model_path: str, port: int) -> None:
         "--dtype", "bfloat16",
         "--max-num-seqs", "64",
         "--max-model-len", "65536",
-        "--max-num-batched-tokens", "32768",
+        # "--max-num-batched-tokens", "32768",
         "--gpu-memory-utilization", "0.8",
-        "--enforce-eager",
+        # "--enforce-eager",
         "--no-enable-prefix-caching",
         "--enable-chunked-prefill",
         "--chat-template-content-format", "openai",