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

[Refactor] Decouple TimingContext from InputProcessingContext (#35083)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-02-23 22:15:50 +08:00
committed by GitHub
parent 1e8438a89a
commit 392645454b
38 changed files with 419 additions and 649 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
tests/models/multimodal/processing/test_tensor_schema.py
View File

@@ -99,7 +99,7 @@ def create_batched_mm_kwargs(
mm_counts=mm_counts,
mm_options={},
)
mm_items = processor_inputs.mm_items
mm_items = processor_inputs.mm_data_items
resized_mm_data = {
modality: resize_mm_data(items.data, size_factors)
for modality, items in mm_items.items()
@@ -108,11 +108,10 @@ def create_batched_mm_kwargs(
# video metadata will be added back to the resized video data here.
text_prompt, token_prompt = get_text_token_prompts(processor, resized_mm_data)
mm_kwargs = processor.apply(
mm_kwargs = processor(
prompt=token_prompt if text_prompt is None else text_prompt,
mm_items=processor.info.parse_mm_data(resized_mm_data),
hf_processor_mm_kwargs=processor_inputs.hf_processor_mm_kwargs,
tokenization_kwargs=processor_inputs.tokenization_kwargs,
)["mm_kwargs"].require_data()
return group_mm_kwargs_by_modality(