more speculative decoding fixes

kimi_k2_tool_parser.py

@@ -57,6 +57,7 @@ from vllm.entrypoints.openai.engine.protocol import (
     FunctionCall,
     ToolCall,
 )
+from vllm.entrypoints.openai.responses.protocol import ResponsesRequest
 from vllm.logger import init_logger
 from vllm.tokenizers import TokenizerLike
 from vllm.tool_parsers.abstract_tool_parser import (
@@ -171,6 +172,20 @@ class KimiK2ToolParser(ToolParser):
             "Successfully initialized %s", self.__class__.__name__
         )
 
+    # ------------------------------------------------------------------
+    # Request adjustment
+    # ------------------------------------------------------------------
+
+    def adjust_request(
+        self, request: ChatCompletionRequest | ResponsesRequest
+    ) -> ChatCompletionRequest | ResponsesRequest:
+        request = super().adjust_request(request)
+        if request.tools and request.tool_choice != "none":
+            # Ensure tool-call tokens (<|tool_calls_section_begin|>,
+            # <|tool_call_begin|>, etc.) are not stripped during decoding.
+            request.skip_special_tokens = False
+        return request
+
     # ------------------------------------------------------------------
     # Helpers
     # ------------------------------------------------------------------