Allow to disable batch decoding. (#11944)

python/sglang/srt/server_args.py

@@ -433,6 +433,7 @@ class ServerArgs:
     enable_symm_mem: bool = False
     disable_flashinfer_cutlass_moe_fp4_allgather: bool = False
     enable_tokenizer_batch_encode: bool = False
+    disable_tokenizer_batch_decode: bool = False
     disable_outlines_disk_cache: bool = False
     disable_custom_all_reduce: bool = False
     enable_mscclpp: bool = False
@@ -2898,6 +2899,11 @@ class ServerArgs:
             action="store_true",
             help="Enable batch tokenization for improved performance when processing multiple text inputs. Do not use with image inputs, pre-tokenized input_ids, or input_embeds.",
         )
+        parser.add_argument(
+            "--disable-tokenizer-batch-decode",
+            action="store_true",
+            help="Disable batch decoding when decoding multiple completions.",
+        )
         parser.add_argument(
             "--disable-outlines-disk-cache",
             action="store_true",