[Misc] Clean up type annotation for SupportsMultiModal (#14794)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-03-14 00:59:56 -07:00
committed by GitHub
parent 09269b3127
commit 601bd3268e
27 changed files with 121 additions and 141 deletions
+4 -5
View File
@@ -30,12 +30,12 @@ from vllm.model_executor.layers.sampler import SamplerOutput, get_sampler
from vllm.model_executor.model_loader.weight_utils import default_weight_loader
from vllm.model_executor.sampling_metadata import SamplingMetadata
from vllm.multimodal import MULTIMODAL_REGISTRY, MultiModalKwargs
from vllm.multimodal.inputs import NestedTensors, PlaceholderRange
from vllm.multimodal.inputs import PlaceholderRange
from vllm.multimodal.utils import consecutive_placeholder_ranges
from vllm.sequence import IntermediateTensors, SequenceData
from vllm.transformers_utils.tokenizer import cached_tokenizer_from_config
from .interfaces import SupportsMultiModal, SupportsPP
from .interfaces import MultiModalEmbeddings, SupportsMultiModal, SupportsPP
from .utils import (init_vllm_registered_model, maybe_prefix,
merge_multimodal_embeddings)
from .vision import VisionEncoderInfo, resolve_visual_encoder_outputs
@@ -221,8 +221,7 @@ class PixtralForConditionalGeneration(nn.Module, SupportsMultiModal,
return get_sampler()
def get_multimodal_embeddings(
self, **kwargs
) -> Union[list[torch.Tensor], torch.Tensor, tuple[torch.Tensor, ...]]:
self, **kwargs: object) -> Optional[MultiModalEmbeddings]:
image_input, image_tokens = self._parse_and_validate_image_input(
**kwargs)
if image_input is None:
@@ -255,7 +254,7 @@ class PixtralForConditionalGeneration(nn.Module, SupportsMultiModal,
def get_input_embeddings(
self,
input_ids: torch.Tensor,
multimodal_embeddings: Optional[NestedTensors] = None,
multimodal_embeddings: Optional[MultiModalEmbeddings] = None,
) -> torch.Tensor:
inputs_embeds = self.language_model.get_input_embeddings(input_ids)
if multimodal_embeddings is not None: