fix comment&log

paddle_inference/paddle/include/paddle_engine.h

@@ -228,7 +228,6 @@ class PaddleInferenceEngine : public EngineCore {
     int local_min_subgraph_size = min_subgraph_size;
     if (engine_conf.has_min_subgraph_size()) {
       local_min_subgraph_size = engine_conf.min_subgraph_size();
-      LOG(INFO) << "local_min_subgraph_size=" << local_min_subgraph_size;
     }
 
     if (engine_conf.has_use_trt() && engine_conf.use_trt()) {

python/paddle_serving_server/serve.py

@@ -223,7 +223,7 @@ def serve_args():
     parser.add_argument(
         "--request_cache_size", type=int, default=0, help="Port of the Prometheus")
     parser.add_argument(
-        "--min_subgraph_size", type=int, default="", nargs="+", help="gpu ids")
+        "--min_subgraph_size", type=int, default="", nargs="+", help="min_subgraph_size")
     return parser.parse_args()