[Misc] Reorganize inputs (#35182)

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

vllm/model_executor/models/transformers/multimodal.py

@@ -22,16 +22,14 @@ from typing import TYPE_CHECKING
 import torch
 
 from vllm.config.utils import getattr_iter
+from vllm.inputs import MultiModalDataDict, MultiModalInput, mm_input
 from vllm.logger import init_logger
 from vllm.model_executor.models.interfaces import SupportsMRoPE, SupportsMultiModal
 from vllm.multimodal import MultiModalKwargsItems
 from vllm.multimodal.inputs import (
-    MultiModalDataDict,
     MultiModalFeatureSpec,
     MultiModalFieldConfig,
-    MultiModalInputs,
     PlaceholderRange,
-    mm_inputs,
 )
 from vllm.multimodal.parse import (
     ImageProcessorItems,
@@ -179,7 +177,7 @@ class MultiModalProcessor(BaseMultiModalProcessor[MultiModalProcessingInfo]):
         self,
         inputs: ProcessorInputs,
         timing_ctx: TimingContext,
-    ) -> MultiModalInputs:
+    ) -> MultiModalInput:
         """
         Process multi-modal inputs to be used in vLLM.
 
@@ -261,7 +259,7 @@ class MultiModalProcessor(BaseMultiModalProcessor[MultiModalProcessingInfo]):
         with timing_ctx.record("get_mm_hashes"):
             mm_hashes = inputs.get_mm_hashes(self.info.model_id)
 
-        return mm_inputs(
+        return mm_input(
             prompt_token_ids=prompt_ids,
             mm_kwargs=mm_kwargs,
             mm_hashes=mm_hashes,