[Refactor] Decouple TimingContext from InputProcessingContext (#35083)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-02-23 14:15:50 +00:00
committed by GitHub
parent 1e8438a89a
commit 392645454b
38 changed files with 419 additions and 649 deletions
+2 -16
View File
@@ -2,7 +2,6 @@
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
from abc import ABC, abstractmethod
from collections.abc import Mapping
from dataclasses import dataclass, field
from typing import Generic, TypeVar
import numpy as np
@@ -18,27 +17,14 @@ from vllm.config.multimodal import (
from vllm.logger import init_logger
from ..inputs import MultiModalDataDict
from ..parse import MultiModalDataItems
from .context import BaseProcessingInfo
from .inputs import ProcessorInputs
_I = TypeVar("_I", bound=BaseProcessingInfo)
logger = init_logger(__name__)
@dataclass
class ProcessorInputs:
"""
Represents the keyword arguments to
[`vllm.multimodal.processing.BaseMultiModalProcessor.apply`][].
"""
prompt: str | list[int]
mm_items: MultiModalDataItems
hf_processor_mm_kwargs: Mapping[str, object] = field(default_factory=dict)
tokenization_kwargs: Mapping[str, object] = field(default_factory=dict)
class BaseDummyInputsBuilder(ABC, Generic[_I]):
"""
Abstract base class that constructs the dummy data to profile
@@ -101,7 +87,7 @@ class BaseDummyInputsBuilder(ABC, Generic[_I]):
return ProcessorInputs(
prompt=dummy_text,
mm_items=dummy_mm_items,
mm_data_items=dummy_mm_items,
tokenization_kwargs=tokenization_kwargs,
)