[MM] Remove text components from ProcessorInputs (#53093)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-08-20 20:42:53 -07:00
committed by GitHub
parent 2adf4b9e2a
commit b00f475f09
92 changed files with 220 additions and 806 deletions
@@ -37,7 +37,7 @@ from vllm.utils.torch_utils import set_default_torch_dtype
from ....utils import create_new_process_for_each_test
from ...registry import HF_EXAMPLE_MODELS
from ...utils import dummy_hf_overrides
from .test_common import get_model_ids_to_test, get_text_token_prompts
from .test_common import get_model_ids_to_test, get_token_prompt
ImageInput = list[Image.Image]
VideoInput: TypeAlias = (
@@ -107,10 +107,10 @@ 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)
token_prompt = get_token_prompt(processor, resized_mm_data)
mm_kwargs = processor(
prompt=token_prompt if text_prompt is None else text_prompt,
prompt=token_prompt,
mm_items=processor.info.parse_mm_data(resized_mm_data),
hf_processor_mm_kwargs=processor_inputs.hf_processor_mm_kwargs,
)["mm_kwargs"].require_data()