biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Refactor] Define MultiModalKwargsItems separate from MultiModalKwargs (#23053)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-08-18 17:52:00 +08:00
committed by GitHub
parent 5c79b0d648
commit 27e8d1ea3e
77 changed files with 431 additions and 383 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
vllm/model_executor/models/qwen2_audio.py
View File

@@ -37,7 +37,7 @@ from vllm.config import VllmConfig
from vllm.model_executor.sampling_metadata import SamplingMetadata
from vllm.multimodal import MULTIMODAL_REGISTRY
from vllm.multimodal.inputs import (MultiModalDataDict, MultiModalFieldConfig,
MultiModalKwargs)
MultiModalKwargsItems)
from vllm.multimodal.parse import (AudioProcessorItems, MultiModalDataItems,
MultiModalDataParser)
from vllm.multimodal.processing import (BaseMultiModalProcessor,
@@ -182,7 +182,7 @@ class Qwen2AudioMultiModalProcessor(
self,
mm_items: MultiModalDataItems,
hf_processor_mm_kwargs: Mapping[str, object],
out_mm_kwargs: MultiModalKwargs,
out_mm_kwargs: MultiModalKwargsItems,
) -> Sequence[PromptUpdate]:
processor = self.info.get_hf_processor(**hf_processor_mm_kwargs)
tokenizer = self.info.get_tokenizer()
@@ -199,7 +199,8 @@ class Qwen2AudioMultiModalProcessor(
audio_bos_id = vocab[audio_bos_token]
audio_eos_id = vocab[audio_eos_token]
feature_attention_mask = out_mm_kwargs.get("feature_attention_mask")
out_mm_data = out_mm_kwargs.get_data()
feature_attention_mask = out_mm_data.get("feature_attention_mask")
if feature_attention_mask is None:
audio_output_lengths = []
else: