[Bugfix] Fix usage of deprecated decorator (#11025)

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

vllm/engine/llm_engine.py

@@ -677,12 +677,10 @@ class LLMEngine:
         self.model_executor.stop_remote_worker_execution_loop()
 
     @overload
-    @deprecated("'inputs' will be renamed to 'prompt")
     def add_request(
         self,
         request_id: str,
-        *,
+        prompt: PromptType,
-        inputs: PromptType,
         params: Union[SamplingParams, PoolingParams],
         arrival_time: Optional[float] = None,
         lora_request: Optional[LoRARequest] = None,
@@ -693,10 +691,12 @@ class LLMEngine:
         ...
 
     @overload
+    @deprecated("'inputs' will be renamed to 'prompt")
     def add_request(
         self,
         request_id: str,
-        prompt: PromptType,
+        *,
+        inputs: PromptType,
         params: Union[SamplingParams, PoolingParams],
         arrival_time: Optional[float] = None,
         lora_request: Optional[LoRARequest] = None,

vllm/engine/multiprocessing/__init__.py

@@ -35,11 +35,9 @@ class RPCProcessRequest:
     priority: int = 0
 
     @overload
-    @deprecated("'inputs' will be renamed to 'prompt")
     def __init__(
         self,
-        *,
+        prompt: PromptType,
-        inputs: PromptType,
         params: Union[SamplingParams, PoolingParams],
         request_id: str,
         lora_request: Optional[LoRARequest] = None,
@@ -50,9 +48,11 @@ class RPCProcessRequest:
         ...
 
     @overload
+    @deprecated("'inputs' will be renamed to 'prompt")
     def __init__(
         self,
-        prompt: PromptType,
+        *,
+        inputs: PromptType,
         params: Union[SamplingParams, PoolingParams],
         request_id: str,
         lora_request: Optional[LoRARequest] = None,

vllm/engine/multiprocessing/client.py

@@ -415,11 +415,9 @@ class MQLLMEngineClient(EngineClient):
         return ENGINE_DEAD_ERROR(self._errored_with)
 
     @overload
-    @deprecated("'inputs' will be renamed to 'prompt")
     def generate(
         self,
-        *,
+        prompt: PromptType,
-        inputs: PromptType,
         sampling_params: SamplingParams,
         request_id: str,
         lora_request: Optional[LoRARequest] = None,
@@ -430,9 +428,11 @@ class MQLLMEngineClient(EngineClient):
         ...
 
     @overload
+    @deprecated("'inputs' will be renamed to 'prompt")
     def generate(
         self,
-        prompt: PromptType,
+        *,
+        inputs: PromptType,
         sampling_params: SamplingParams,
         request_id: str,
         lora_request: Optional[LoRARequest] = None,
@@ -487,11 +487,9 @@ class MQLLMEngineClient(EngineClient):
                                      prompt_adapter_request, priority)
 
     @overload
-    @deprecated("'inputs' will be renamed to 'prompt")
     def encode(
         self,
-        *,
+        prompt: PromptType,
-        inputs: PromptType,
         pooling_params: PoolingParams,
         request_id: str,
         lora_request: Optional[LoRARequest] = None,
@@ -501,9 +499,11 @@ class MQLLMEngineClient(EngineClient):
         ...
 
     @overload
+    @deprecated("'inputs' will be renamed to 'prompt")
     def encode(
         self,
-        prompt: PromptType,
+        *,
+        inputs: PromptType,
         pooling_params: PoolingParams,
         request_id: str,
         lora_request: Optional[LoRARequest] = None,

vllm/entrypoints/llm.py

@@ -252,8 +252,21 @@ class LLM:
         else:
             tokenizer_group.tokenizer = get_cached_tokenizer(tokenizer)
 
+    @overload
+    def generate(
+        self,
+        prompts: Union[PromptType, Sequence[PromptType]],
+        /,
+        *,
+        sampling_params: Optional[Union[SamplingParams,
+                                        Sequence[SamplingParams]]] = None,
+        use_tqdm: bool = True,
+        lora_request: Optional[Union[List[LoRARequest], LoRARequest]] = None,
+    ) -> List[RequestOutput]:
+        ...
+
     @overload  # LEGACY: single (prompt + optional token ids)
-    @deprecated("'prompt_token_ids' will become part of 'prompts")
+    @deprecated("'prompt_token_ids' will become part of 'prompts'")
     def generate(
         self,
         prompts: str,
@@ -266,7 +279,7 @@ class LLM:
         ...
 
     @overload  # LEGACY: multi (prompt + optional token ids)
-    @deprecated("'prompt_token_ids' will become part of 'prompts")
+    @deprecated("'prompt_token_ids' will become part of 'prompts'")
     def generate(
         self,
         prompts: List[str],
@@ -279,7 +292,7 @@ class LLM:
         ...
 
     @overload  # LEGACY: single (token ids + optional prompt)
-    @deprecated("'prompt_token_ids' will become part of 'prompts")
+    @deprecated("'prompt_token_ids' will become part of 'prompts'")
     def generate(
         self,
         prompts: Optional[str] = None,
@@ -293,7 +306,7 @@ class LLM:
         ...
 
     @overload  # LEGACY: multi (token ids + optional prompt)
-    @deprecated("'prompt_token_ids' will become part of 'prompts")
+    @deprecated("'prompt_token_ids' will become part of 'prompts'")
     def generate(
         self,
         prompts: Optional[List[str]] = None,
@@ -307,7 +320,7 @@ class LLM:
         ...
 
     @overload  # LEGACY: single or multi token ids [pos-only]
-    @deprecated("'prompt_token_ids' will become part of 'prompts")
+    @deprecated("'prompt_token_ids' will become part of 'prompts'")
     def generate(
         self,
         prompts: None,
@@ -318,19 +331,6 @@ class LLM:
     ) -> List[RequestOutput]:
         ...
 
-    @overload
-    def generate(
-        self,
-        prompts: Union[PromptType, Sequence[PromptType]],
-        /,
-        *,
-        sampling_params: Optional[Union[SamplingParams,
-                                        Sequence[SamplingParams]]] = None,
-        use_tqdm: bool = True,
-        lora_request: Optional[Union[List[LoRARequest], LoRARequest]] = None,
-    ) -> List[RequestOutput]:
-        ...
-
     @deprecate_kwargs(
         "prompt_token_ids",
         is_deprecated=lambda: LLM.DEPRECATE_LEGACY,
@@ -672,8 +672,21 @@ class LLM:
             lora_request=lora_request,
         )
 
+    @overload
+    def encode(
+        self,
+        prompts: Union[PromptType, Sequence[PromptType]],
+        /,
+        *,
+        pooling_params: Optional[Union[PoolingParams,
+                                       Sequence[PoolingParams]]] = None,
+        use_tqdm: bool = True,
+        lora_request: Optional[Union[List[LoRARequest], LoRARequest]] = None,
+    ) -> List[PoolingRequestOutput]:
+        ...
+
     @overload  # LEGACY: single (prompt + optional token ids)
-    @deprecated("'prompt_token_ids' will become part of 'prompts")
+    @deprecated("'prompt_token_ids' will become part of 'prompts'")
     def encode(
         self,
         prompts: str,
@@ -686,7 +699,7 @@ class LLM:
         ...
 
     @overload  # LEGACY: multi (prompt + optional token ids)
-    @deprecated("'prompt_token_ids' will become part of 'prompts")
+    @deprecated("'prompt_token_ids' will become part of 'prompts'")
     def encode(
         self,
         prompts: List[str],
@@ -699,7 +712,7 @@ class LLM:
         ...
 
     @overload  # LEGACY: single (token ids + optional prompt)
-    @deprecated("'prompt_token_ids' will become part of 'prompts")
+    @deprecated("'prompt_token_ids' will become part of 'prompts'")
     def encode(
         self,
         prompts: Optional[str] = None,
@@ -713,7 +726,7 @@ class LLM:
         ...
 
     @overload  # LEGACY: multi (token ids + optional prompt)
-    @deprecated("'prompt_token_ids' will become part of 'prompts")
+    @deprecated("'prompt_token_ids' will become part of 'prompts'")
     def encode(
         self,
         prompts: Optional[List[str]] = None,
@@ -727,7 +740,7 @@ class LLM:
         ...
 
     @overload  # LEGACY: single or multi token ids [pos-only]
-    @deprecated("'prompt_token_ids' will become part of 'prompts")
+    @deprecated("'prompt_token_ids' will become part of 'prompts'")
     def encode(
         self,
         prompts: None,
@@ -738,19 +751,6 @@ class LLM:
     ) -> List[PoolingRequestOutput]:
         ...
 
-    @overload
-    def encode(
-        self,
-        prompts: Union[PromptType, Sequence[PromptType]],
-        /,
-        *,
-        pooling_params: Optional[Union[PoolingParams,
-                                       Sequence[PoolingParams]]] = None,
-        use_tqdm: bool = True,
-        lora_request: Optional[Union[List[LoRARequest], LoRARequest]] = None,
-    ) -> List[PoolingRequestOutput]:
-        ...
-
     @deprecate_kwargs(
         "prompt_token_ids",
         is_deprecated=lambda: LLM.DEPRECATE_LEGACY,