[Spec][Ngram] 4/N: Remove max_match_window_size and min_match_window_size, matching all suffixes of the Trie (#21225)

python/sglang/srt/server_args.py

@@ -506,8 +506,6 @@ class ServerArgs:
     speculative_draft_model_quantization: Optional[str] = None
 
     # Speculative decoding (ngram)
-    speculative_ngram_min_match_window_size: int = 1
-    speculative_ngram_max_match_window_size: int = 12
     speculative_ngram_min_bfs_breadth: int = 1
     speculative_ngram_max_bfs_breadth: int = 10
     speculative_ngram_match_type: Literal["BFS", "PROB"] = "BFS"
@@ -3108,8 +3106,10 @@ class ServerArgs:
             self.enable_mixed_chunk = False
             self.speculative_eagle_topk = self.speculative_ngram_max_bfs_breadth
             if self.speculative_num_draft_tokens is None:
-                self.speculative_num_draft_tokens = (
-                    self.speculative_ngram_max_match_window_size
+                self.speculative_num_draft_tokens = 12
+                logger.warning(
+                    "speculative_num_draft_tokens is set to 12 by default for ngram speculative decoding. "
+                    "You can override this by explicitly setting --speculative-num-draft-tokens."
                 )
             logger.warning(
                 "The overlap scheduler and mixed chunked prefill are disabled because of "
@@ -4851,18 +4851,6 @@ class ServerArgs:
         )
 
         # Speculative decoding (ngram)
-        parser.add_argument(
-            "--speculative-ngram-min-match-window-size",
-            type=int,
-            default=ServerArgs.speculative_ngram_min_match_window_size,
-            help="The minimum window size for pattern matching in ngram speculative decoding.",
-        )
-        parser.add_argument(
-            "--speculative-ngram-max-match-window-size",
-            type=int,
-            default=ServerArgs.speculative_ngram_max_match_window_size,
-            help="The maximum window size for pattern matching in ngram speculative decoding.",
-        )
         parser.add_argument(
             "--speculative-ngram-min-bfs-breadth",
             type=int,