[Misc] Remove deprecated names

vllm/engine/async_llm_engine.py

@@ -6,6 +6,8 @@
                     List, Mapping, Optional, Set, Tuple, Type, Union, overload)
 from weakref import ReferenceType
 
+from typing_extensions import deprecated
+
 import vllm.envs as envs
 from vllm.config import (DecodingConfig, LoRAConfig, ModelConfig,
                          ParallelConfig, SchedulerConfig, VllmConfig)
@@ -422,7 +424,8 @@ async def get_tokenizer_async(self,
         return await (
             self.get_tokenizer_group().get_lora_tokenizer_async(lora_request))
 
-    @overload  # DEPRECATED
+    @overload
+    @deprecated("'inputs' will be renamed to 'prompt")
     async def add_request_async(
         self,
         request_id: str,
@@ -894,7 +897,8 @@ async def run_engine_loop(engine_ref: ReferenceType):
 
     # This method does not need to be async, but kept that way
     # for backwards compatibility.
-    @overload  # DEPRECATED
+    @overload
+    @deprecated("'inputs' will be renamed to 'prompt")
     def add_request(
         self,
         request_id: str,

vllm/engine/llm_engine.py

@@ -10,7 +10,7 @@
 from typing import Set, Type, Union, cast, overload
 
 import torch
-from typing_extensions import TypeVar
+from typing_extensions import TypeVar, deprecated
 
 import vllm.envs as envs
 from vllm.config import (DecodingConfig, LoRAConfig, ModelConfig,
@@ -719,7 +719,8 @@ def _add_processed_request(
     def stop_remote_worker_execution_loop(self) -> None:
         self.model_executor.stop_remote_worker_execution_loop()
 
-    @overload  # DEPRECATED
+    @overload
+    @deprecated("'inputs' will be renamed to 'prompt")
     def add_request(
         self,
         request_id: str,

vllm/engine/multiprocessing/__init__.py

@@ -2,6 +2,8 @@
 from enum import Enum
 from typing import List, Mapping, Optional, Union, overload
 
+from typing_extensions import deprecated
+
 from vllm import PoolingParams
 from vllm.inputs import PromptType
 from vllm.lora.request import LoRARequest
@@ -32,7 +34,8 @@ class RPCProcessRequest:
     prompt_adapter_request: Optional[PromptAdapterRequest] = None
     priority: int = 0
 
-    @overload  # DEPRECATED
+    @overload
+    @deprecated("'inputs' will be renamed to 'prompt")
     def __init__(
         self,
         *,

vllm/engine/multiprocessing/client.py

@@ -9,6 +9,7 @@
 import psutil
 import zmq
 import zmq.asyncio
+from typing_extensions import deprecated
 from zmq import Frame  # type: ignore[attr-defined]
 from zmq.asyncio import Socket
 
@@ -414,7 +415,8 @@ def errored(self) -> bool:
     def dead_error(self) -> BaseException:
         return ENGINE_DEAD_ERROR(self._errored_with)
 
-    @overload  # DEPRECATED
+    @overload
+    @deprecated("'inputs' will be renamed to 'prompt")
     def generate(
         self,
         *,
@@ -485,7 +487,8 @@ def generate(
                                      lora_request, trace_headers,
                                      prompt_adapter_request, priority)
 
-    @overload  # DEPRECATED
+    @overload
+    @deprecated("'inputs' will be renamed to 'prompt")
     def encode(
         self,
         *,

vllm/entrypoints/llm.py

@@ -6,6 +6,7 @@
                     Union, cast, overload)
 
 from tqdm import tqdm
+from typing_extensions import deprecated
 
 from vllm import envs
 from vllm.beam_search import (BeamSearchInstance, BeamSearchOutput,
@@ -256,6 +257,7 @@ def set_tokenizer(self, tokenizer: AnyTokenizer) -> None:
             tokenizer_group.tokenizer = get_cached_tokenizer(tokenizer)
 
     @overload  # LEGACY: single (prompt + optional token ids)
+    @deprecated("'prompt_token_ids' will become part of 'prompts")
     def generate(
         self,
         prompts: str,
@@ -268,6 +270,7 @@ def generate(
         ...
 
     @overload  # LEGACY: multi (prompt + optional token ids)
+    @deprecated("'prompt_token_ids' will become part of 'prompts")
     def generate(
         self,
         prompts: List[str],
@@ -280,6 +283,7 @@ def generate(
         ...
 
     @overload  # LEGACY: single (token ids + optional prompt)
+    @deprecated("'prompt_token_ids' will become part of 'prompts")
     def generate(
         self,
         prompts: Optional[str] = None,
@@ -293,6 +297,7 @@ def generate(
         ...
 
     @overload  # LEGACY: multi (token ids + optional prompt)
+    @deprecated("'prompt_token_ids' will become part of 'prompts")
     def generate(
         self,
         prompts: Optional[List[str]] = None,
@@ -306,6 +311,7 @@ def generate(
         ...
 
     @overload  # LEGACY: single or multi token ids [pos-only]
+    @deprecated("'prompt_token_ids' will become part of 'prompts")
     def generate(
         self,
         prompts: None,
@@ -671,6 +677,7 @@ def chat(
         )
 
     @overload  # LEGACY: single (prompt + optional token ids)
+    @deprecated("'prompt_token_ids' will become part of 'prompts")
     def encode(
         self,
         prompts: str,
@@ -683,6 +690,7 @@ def encode(
         ...
 
     @overload  # LEGACY: multi (prompt + optional token ids)
+    @deprecated("'prompt_token_ids' will become part of 'prompts")
     def encode(
         self,
         prompts: List[str],
@@ -695,6 +703,7 @@ def encode(
         ...
 
     @overload  # LEGACY: single (token ids + optional prompt)
+    @deprecated("'prompt_token_ids' will become part of 'prompts")
     def encode(
         self,
         prompts: Optional[str] = None,
@@ -708,6 +717,7 @@ def encode(
         ...
 
     @overload  # LEGACY: multi (token ids + optional prompt)
+    @deprecated("'prompt_token_ids' will become part of 'prompts")
     def encode(
         self,
         prompts: Optional[List[str]] = None,
@@ -721,6 +731,7 @@ def encode(
         ...
 
     @overload  # LEGACY: single or multi token ids [pos-only]
+    @deprecated("'prompt_token_ids' will become part of 'prompts")
     def encode(
         self,
         prompts: None,

vllm/inputs/__init__.py

@@ -38,34 +38,3 @@
     "InputProcessingContext",
     "InputRegistry",
 ]
-
-
-def __getattr__(name: str):
-    import warnings
-
-    if name == "PromptInput":
-        msg = ("PromptInput has been renamed to PromptType. "
-               "The original name will be removed in an upcoming version.")
-
-        warnings.warn(DeprecationWarning(msg), stacklevel=2)
-
-        return PromptType
-
-    if name == "LLMInputs":
-        msg = ("LLMInputs has been renamed to DecoderOnlyInputs. "
-               "The original name will be removed in an upcoming version.")
-
-        warnings.warn(DeprecationWarning(msg), stacklevel=2)
-
-        return DecoderOnlyInputs
-
-    if name == "EncoderDecoderLLMInputs":
-        msg = (
-            "EncoderDecoderLLMInputs has been renamed to EncoderDecoderInputs. "
-            "The original name will be removed in an upcoming version.")
-
-        warnings.warn(DeprecationWarning(msg), stacklevel=2)
-
-        return EncoderDecoderInputs
-
-    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

vllm/inputs/data.py

@@ -358,34 +358,3 @@ def to_enc_dec_tuple_list(
     return [(enc_dec_prompt["encoder_prompt"],
              enc_dec_prompt["decoder_prompt"])
             for enc_dec_prompt in enc_dec_prompts]
-
-
-def __getattr__(name: str):
-    import warnings
-
-    if name == "PromptInput":
-        msg = ("PromptInput has been renamed to PromptType. "
-               "The original name will be removed in an upcoming version.")
-
-        warnings.warn(DeprecationWarning(msg), stacklevel=2)
-
-        return PromptType
-
-    if name == "LLMInputs":
-        msg = ("LLMInputs has been renamed to DecoderOnlyInputs. "
-               "The original name will be removed in an upcoming version.")
-
-        warnings.warn(DeprecationWarning(msg), stacklevel=2)
-
-        return DecoderOnlyInputs
-
-    if name == "EncoderDecoderLLMInputs":
-        msg = (
-            "EncoderDecoderLLMInputs has been renamed to EncoderDecoderInputs. "
-            "The original name will be removed in an upcoming version.")
-
-        warnings.warn(DeprecationWarning(msg), stacklevel=2)
-
-        return EncoderDecoderInputs
-
-    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

vllm/model_executor/models/aria.py

@@ -32,9 +32,8 @@
                                               maybe_prefix,
                                               merge_multimodal_embeddings)
 from vllm.multimodal import MULTIMODAL_REGISTRY
-from vllm.multimodal.base import MultiModalInputs
 from vllm.multimodal.image import cached_get_image_processor
-from vllm.multimodal.inputs import NestedTensors
+from vllm.multimodal.inputs import MultiModalKwargs, NestedTensors
 from vllm.multimodal.utils import (cached_get_tokenizer,
                                    repeat_and_pad_placeholder_tokens)
 from vllm.sequence import IntermediateTensors
@@ -451,7 +450,7 @@ def get_max_multimodal_tokens(ctx):
 
 
 def input_mapper_for_aria(ctx, data):
-    return MultiModalInputs(data)
+    return MultiModalKwargs(data)
 
 
 def input_processor(ctx, llm_inputs):

vllm/multimodal/__init__.py

@@ -27,18 +27,3 @@
     "MULTIMODAL_REGISTRY",
     "MultiModalRegistry",
 ]
-
-
-def __getattr__(name: str):
-    import warnings
-
-    if name == "MultiModalInputs":
-        msg = ("MultiModalInputs has been renamed to MultiModalKwargs. "
-               "The original name will take another meaning in an upcoming "
-               "version.")
-
-        warnings.warn(DeprecationWarning(msg), stacklevel=2)
-
-        return MultiModalKwargs
-
-    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

vllm/multimodal/base.py

@@ -433,18 +433,3 @@ def index_map(self) -> "IndexMap":
 
         return MultiModalPlaceholderMap.IndexMap(src=src_indices,
                                                  dest=dest_indices)
-
-
-def __getattr__(name: str):
-    import warnings
-
-    if name == "MultiModalInputs":
-        msg = ("MultiModalInputs has been renamed to MultiModalKwargs. "
-               "The original name will take another meaning in an upcoming "
-               "version.")
-
-        warnings.warn(DeprecationWarning(msg), stacklevel=2)
-
-        return MultiModalKwargs
-
-    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")