Skip to content

[Deprecation] Remove inputs arg fallback in Engine classes #18799

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 4 commits into from
Jun 9, 2025
Merged

[Deprecation] Remove inputs arg fallback in Engine classes #18799

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
77d694c
[Deprecation] Remove `inputs` arg fallback
DarkLight1337 May 28, 2025
83f2765
Merge branch 'main' into deprecate-inputs-arg
DarkLight1337 Jun 4, 2025
b8769ba
Merge branch 'main' into deprecate-inputs-arg
DarkLight1337 Jun 6, 2025
0f99f9b
Merge branch 'main' into deprecate-inputs-arg
DarkLight1337 Jun 8, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
107 changes: 9 additions & 98 deletions vllm/engine/async_llm_engine.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,12 +6,10 @@
import time
import weakref
from functools import partial
from typing import (Any, AsyncGenerator, Callable, Coroutine, Dict, Iterable,
List, Mapping, Optional, Set, Tuple, Type, Union, overload)
from typing import (Any, AsyncGenerator, Callable, Dict, Iterable, List,
Mapping, Optional, Set, Tuple, Type, Union)
from weakref import ReferenceType

from typing_extensions import deprecated

import vllm.envs as envs
from vllm.config import (DecodingConfig, LoRAConfig, ModelConfig,
ParallelConfig, SchedulerConfig, VllmConfig)
Expand All @@ -36,7 +34,7 @@
from vllm.sequence import ExecuteModelRequest
from vllm.transformers_utils.tokenizer import AnyTokenizer
from vllm.usage.usage_lib import UsageContext
from vllm.utils import Device, deprecate_kwargs, weak_bind
from vllm.utils import Device, weak_bind

logger = init_logger(__name__)
ENGINE_ITERATION_TIMEOUT_S = envs.VLLM_ENGINE_ITERATION_TIMEOUT_S
Expand Down Expand Up @@ -429,24 +427,6 @@ async def get_tokenizer_async(self,
return await (
self.get_tokenizer_group().get_lora_tokenizer_async(lora_request))

@overload
@deprecated("'inputs' will be renamed to 'prompt")
async def add_request_async(
self,
request_id: str,
*,
inputs: PromptType,
params: Union[SamplingParams, PoolingParams],
arrival_time: Optional[float] = None,
lora_request: Optional[LoRARequest] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
data_parallel_rank: Optional[int] = None,
) -> None:
...

@overload
async def add_request_async(
self,
request_id: str,
Expand All @@ -459,32 +439,10 @@ async def add_request_async(
priority: int = 0,
data_parallel_rank: Optional[int] = None,
) -> None:
...

@deprecate_kwargs(
"inputs",
additional_message="Please use the 'prompt' parameter instead.",
)
async def add_request_async(
self,
request_id: str,
prompt: Optional[PromptType] = None,
params: Optional[Union[SamplingParams, PoolingParams]] = None,
arrival_time: Optional[float] = None,
lora_request: Optional[LoRARequest] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
data_parallel_rank: Optional[int] = None,
*,
inputs: Optional[PromptType] = None, # DEPRECATED
) -> None:
"""Async version of
[`add_request`][vllm.engine.llm_engine.LLMEngine.add_request]."""
if inputs is not None:
prompt = inputs
assert prompt is not None and params is not None

"""
Async version of
[`add_request`][vllm.engine.llm_engine.LLMEngine.add_request].
"""
if lora_request is not None and not self.lora_config:
raise ValueError(f"Got lora_request {lora_request} but LoRA is "
"not enabled!")
Expand Down Expand Up @@ -521,8 +479,7 @@ async def add_request_async(
params = await build_guided_decoding_logits_processor_async(
sampling_params=params,
tokenizer=await self.get_tokenizer_async(lora_request),
default_guided_backend=self.decoding_config.
guided_decoding_backend,
default_guided_backend=self.decoding_config.backend,
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Resolves a deprecation from #17008

reasoning_backend=self.decoding_config.reasoning_backend,
model_config=self.model_config)

Expand Down Expand Up @@ -894,28 +851,7 @@ async def run_engine_loop(engine_ref: ReferenceType):
raise
await asyncio.sleep(0)

# This method does not need to be async, but kept that way
# for backwards compatibility.
@overload
@deprecated("'inputs' will be renamed to 'prompt")
def add_request(
self,
request_id: str,
*,
inputs: PromptType,
params: Union[SamplingParams, PoolingParams],
arrival_time: Optional[float] = None,
lora_request: Optional[LoRARequest] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
data_parallel_rank: Optional[int] = None,
) -> Coroutine[None, None, AsyncGenerator[Union[
RequestOutput, PoolingRequestOutput], None]]:
...

@overload
def add_request(
async def add_request(
self,
request_id: str,
prompt: PromptType,
Expand All @@ -926,32 +862,7 @@ def add_request(
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
data_parallel_rank: Optional[int] = None,
) -> Coroutine[None, None, AsyncGenerator[Union[
RequestOutput, PoolingRequestOutput], None]]:
...

@deprecate_kwargs(
"inputs",
additional_message="Please use the 'prompt' parameter instead.",
)
async def add_request(
self,
request_id: str,
prompt: Optional[PromptType] = None,
params: Optional[Union[SamplingParams, PoolingParams]] = None,
arrival_time: Optional[float] = None,
lora_request: Optional[LoRARequest] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
data_parallel_rank: Optional[int] = None,
*,
inputs: Optional[PromptType] = None, # DEPRECATED
) -> AsyncGenerator[Union[RequestOutput, PoolingRequestOutput], None]:
if inputs is not None:
prompt = inputs
assert prompt is not None and params is not None

if not self.is_running:
if self.start_engine_loop:
self.start_background_loop()
Expand Down
48 changes: 3 additions & 45 deletions vllm/engine/llm_engine.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,10 +11,10 @@
from typing import (TYPE_CHECKING, Any, Callable, ClassVar, Deque, Dict,
Iterable, List, Literal, Mapping, NamedTuple, Optional)
from typing import Sequence as GenericSequence
from typing import Set, Type, Union, cast, overload
from typing import Set, Type, Union, cast

import torch
from typing_extensions import TypeVar, deprecated
from typing_extensions import TypeVar

import vllm.envs as envs
from vllm.config import (DecodingConfig, LoRAConfig, ModelConfig,
Expand Down Expand Up @@ -58,8 +58,7 @@
TokenizerGroup, init_tokenizer_from_configs)
from vllm.usage.usage_lib import (UsageContext, is_usage_stats_enabled,
usage_message)
from vllm.utils import (Counter, Device, deprecate_kwargs,
resolve_obj_by_qualname, weak_bind)
from vllm.utils import Counter, Device, resolve_obj_by_qualname, weak_bind
from vllm.version import __version__ as VLLM_VERSION
from vllm.worker.model_runner_base import InputProcessingError

Expand Down Expand Up @@ -629,7 +628,6 @@ def _add_processed_request(
def stop_remote_worker_execution_loop(self) -> None:
self.model_executor.stop_remote_worker_execution_loop()

@overload
def add_request(
self,
request_id: str,
Expand All @@ -641,42 +639,6 @@ def add_request(
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
) -> None:
...

@overload
@deprecated("'inputs' will be renamed to 'prompt")
def add_request(
self,
request_id: str,
*,
inputs: PromptType,
params: Union[SamplingParams, PoolingParams],
arrival_time: Optional[float] = None,
lora_request: Optional[LoRARequest] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
) -> None:
...

@deprecate_kwargs(
"inputs",
additional_message="Please use the 'prompt' parameter instead.",
)
def add_request(
self,
request_id: str,
prompt: Optional[PromptType] = None,
params: Optional[Union[SamplingParams, PoolingParams]] = None,
arrival_time: Optional[float] = None,
lora_request: Optional[LoRARequest] = None,
tokenization_kwargs: Optional[dict[str, Any]] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
*,
inputs: Optional[PromptType] = None, # DEPRECATED
) -> None:
"""Add a request to the engine's request pool.

Expand Down Expand Up @@ -725,10 +687,6 @@ def add_request(
>>> # continue the request processing
>>> ...
"""
if inputs is not None:
prompt = inputs
assert prompt is not None and params is not None

if lora_request is not None and not self.lora_config:
raise ValueError(f"Got lora_request {lora_request} but LoRA is "
"not enabled!")
Expand Down
45 changes: 2 additions & 43 deletions vllm/engine/multiprocessing/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,17 +4,15 @@
import uuid
from dataclasses import dataclass, field
from enum import Enum
from typing import List, Mapping, Optional, Union, overload

from typing_extensions import deprecated
from typing import List, Mapping, Optional, Union

from vllm import PoolingParams
from vllm.inputs import PromptType
from vllm.lora.request import LoRARequest
from vllm.outputs import RequestOutput
from vllm.prompt_adapter.request import PromptAdapterRequest
from vllm.sampling_params import SamplingParams
from vllm.utils import Device, deprecate_kwargs
from vllm.utils import Device

VLLM_RPC_SUCCESS_STR = "SUCCESS"

Expand All @@ -38,7 +36,6 @@ class RPCProcessRequest:
prompt_adapter_request: Optional[PromptAdapterRequest] = None
priority: int = 0

@overload
def __init__(
self,
prompt: PromptType,
Expand All @@ -49,44 +46,6 @@ def __init__(
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
) -> None:
...

@overload
@deprecated("'inputs' will be renamed to 'prompt")
def __init__(
self,
*,
inputs: PromptType,
params: Union[SamplingParams, PoolingParams],
request_id: str,
lora_request: Optional[LoRARequest] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
) -> None:
...

@deprecate_kwargs(
"inputs",
additional_message="Please use the 'prompt' parameter instead.",
)
def __init__(
self,
prompt: Optional[PromptType] = None,
params: Optional[Union[SamplingParams, PoolingParams]] = None,
request_id: Optional[str] = None,
lora_request: Optional[LoRARequest] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
*,
inputs: Optional[PromptType] = None, # DEPRECATED
) -> None:
if inputs is not None:
prompt = inputs
assert (prompt is not None and params is not None
and request_id is not None)

super().__init__()

self.prompt = prompt
Expand Down
Loading