[Chore] Remove redundant RequestPrompt (#30612)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>

vllm/entrypoints/serve/disagg/serving.py

@@ -27,7 +27,7 @@ from vllm.entrypoints.serve.disagg.protocol import (
     GenerateResponse,
     GenerateResponseChoice,
 )
-from vllm.inputs.data import TokensPrompt as EngineTokensPrompt
+from vllm.inputs.data import TokensPrompt
 from vllm.logger import init_logger
 from vllm.logprobs import Logprob
 from vllm.outputs import RequestOutput
@@ -99,7 +99,7 @@ class ServingTokens(OpenAIServing):
 
         # TODO(NickLucche): Change to EngineCoreRequest once Renderer work is
         # completed
-        engine_prompt = EngineTokensPrompt(prompt_token_ids=request.token_ids)
+        engine_prompt = TokensPrompt(prompt_token_ids=request.token_ids)
         if request.features is not None:
             engine_prompt["multi_modal_data"] = None
 
@@ -115,7 +115,7 @@ class ServingTokens(OpenAIServing):
 
             self._log_inputs(
                 request_id,
-                request.token_ids,
+                TokensPrompt(prompt_token_ids=request.token_ids),
                 params=sampling_params,
                 lora_request=lora_request,
             )

vllm/entrypoints/serve/tokenize/serving.py

@@ -21,6 +21,7 @@ from vllm.entrypoints.openai.protocol import (
 from vllm.entrypoints.openai.serving_engine import OpenAIServing
 from vllm.entrypoints.openai.serving_models import OpenAIServingModels
 from vllm.entrypoints.renderer import RenderConfig
+from vllm.inputs import TokensPrompt
 from vllm.logger import init_logger
 from vllm.tokenizers import TokenizerLike
 
@@ -80,11 +81,8 @@ class OpenAIServingTokenization(OpenAIServing):
                 )
                 if error_check_ret is not None:
                     return error_check_ret
-                (
-                    _,
-                    _,
-                    engine_prompts,
-                ) = await self._preprocess_chat(
+
+                _, engine_prompts = await self._preprocess_chat(
                     request,
                     tokenizer,
                     request.messages,
@@ -141,7 +139,10 @@ class OpenAIServingTokenization(OpenAIServing):
         tokenizer = await self.engine_client.get_tokenizer()
 
         self._log_inputs(
-            request_id, request.tokens, params=None, lora_request=lora_request
+            request_id,
+            TokensPrompt(prompt_token_ids=request.tokens),
+            params=None,
+            lora_request=lora_request,
         )
 
         prompt_input = await self._tokenize_prompt_input_async(