[core] add extra_args to RPCProcessRequest and MQLLMEngineClient

vllm/engine/multiprocessing/__init__.py

@@ -3,7 +3,7 @@
 import uuid
 from dataclasses import dataclass, field
 from enum import Enum
-from typing import List, Mapping, Optional, Union, overload
+from typing import Any, List, Mapping, Optional, Union, overload
 
 from typing_extensions import deprecated
 
@@ -36,6 +36,7 @@ class RPCProcessRequest:
     trace_headers: Optional[Mapping[str, str]] = None
     prompt_adapter_request: Optional[PromptAdapterRequest] = None
     priority: int = 0
+    extra_args: Optional[Mapping[str, Any]] = None
 
     @overload
     def __init__(
@@ -47,6 +48,7 @@ def __init__(
         trace_headers: Optional[Mapping[str, str]] = None,
         prompt_adapter_request: Optional[PromptAdapterRequest] = None,
         priority: int = 0,
+        extra_args: Optional[Mapping[str, Any]] = None,
     ) -> None:
         ...
 
@@ -62,6 +64,7 @@ def __init__(
         trace_headers: Optional[Mapping[str, str]] = None,
         prompt_adapter_request: Optional[PromptAdapterRequest] = None,
         priority: int = 0,
+        extra_args: Optional[Mapping[str, Any]] = None,
     ) -> None:
         ...
 
@@ -78,6 +81,7 @@ def __init__(
             trace_headers: Optional[Mapping[str, str]] = None,
             prompt_adapter_request: Optional[PromptAdapterRequest] = None,
             priority: int = 0,
+            extra_args: Optional[Mapping[str, Any]] = None,
             *,
             inputs: Optional[PromptType] = None,  # DEPRECATED
     ) -> None:
@@ -95,6 +99,7 @@ def __init__(
         self.trace_headers = trace_headers
         self.prompt_adapter_request = prompt_adapter_request
         self.priority = priority
+        self.extra_args = extra_args
 
 
 @dataclass

vllm/engine/multiprocessing/client.py

@@ -442,6 +442,7 @@ def generate(
         trace_headers: Optional[Mapping[str, str]] = None,
         prompt_adapter_request: Optional[PromptAdapterRequest] = None,
         priority: int = 0,
+        extra_args: Optional[Mapping[str, Any]] = None,        
     ) -> AsyncGenerator[RequestOutput, None]:
         ...
 
@@ -457,6 +458,7 @@ def generate(
         trace_headers: Optional[Mapping[str, str]] = None,
         prompt_adapter_request: Optional[PromptAdapterRequest] = None,
         priority: int = 0,
+        extra_args: Optional[Mapping[str, Any]] = None,        
     ) -> AsyncGenerator[RequestOutput, None]:
         ...
 
@@ -473,6 +475,7 @@ def generate(
         trace_headers: Optional[Mapping[str, str]] = None,
         prompt_adapter_request: Optional[PromptAdapterRequest] = None,
         priority: int = 0,
+        extra_args: Optional[Mapping[str, Any]] = None,        
         *,
         inputs: Optional[PromptType] = None  # DEPRECATED
     ) -> AsyncGenerator[RequestOutput, None]:
@@ -502,7 +505,7 @@ def generate(
 
         return self._process_request(prompt, sampling_params, request_id,
                                      lora_request, trace_headers,
-                                     prompt_adapter_request, priority)
+                                     prompt_adapter_request, priority, extra_args)
 
     @overload
     def encode(
@@ -586,6 +589,7 @@ async def _process_request(
         trace_headers: Optional[Mapping[str, str]] = None,
         prompt_adapter_request: Optional[PromptAdapterRequest] = None,
         priority: int = 0,
+        extra_args: Optional[Mapping[str, Any]] = None,        
     ) -> Union[AsyncGenerator[RequestOutput, None], AsyncGenerator[
             PoolingRequestOutput, None]]:
         """Send an RPCGenerateRequest to the RPCServer and stream responses."""
@@ -639,6 +643,7 @@ async def _process_request(
                     trace_headers=trace_headers,
                     prompt_adapter_request=prompt_adapter_request,
                     priority=priority,
+                    extra_args=extra_args,
                 ))
 
             # 3) Send the RPCGenerateRequest to the MQLLMEngine.