[Refactor] Remove InputPreprocessor (#53064)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-08-20 08:39:48 +00:00
committed by GitHub
parent 14617c2b6c
commit 4666a8ba9e
8 changed files with 29 additions and 367 deletions
@@ -54,11 +54,10 @@ def test_text_only_model_mm_data_maps_to_bad_request():
with pytest.raises(ValueError, match="text-only") as exc_info:
renderer._process_multimodal(
prompt="What is in this image?",
prompt=[1],
mm_data={"image": [cherry_pil_image]},
mm_uuids=None,
mm_processor_kwargs=None,
tokenization_kwargs=None,
)
error_response = create_error_response(exc_info.value)
-31
View File
@@ -1,31 +0,0 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
import pytest
from vllm.config import ModelConfig, VllmConfig
from vllm.inputs.preprocess import InputPreprocessor
pytestmark = pytest.mark.cpu_test
@pytest.mark.parametrize("model_id", ["facebook/chameleon-7b"])
@pytest.mark.parametrize("prompt", ["", {"prompt_token_ids": []}])
@pytest.mark.skip(
reason=(
"Applying huggingface processor on text inputs results in "
"significant performance regression for multimodal models. "
"See https://github.com/vllm-project/vllm/issues/26320"
)
)
def test_preprocessor_always_mm_code_path(model_id, prompt):
model_config = ModelConfig(model=model_id)
vllm_config = VllmConfig(model_config=model_config)
input_preprocessor = InputPreprocessor(vllm_config)
# HF processor adds sep token
tokenizer = input_preprocessor.get_tokenizer()
sep_token_id = tokenizer.vocab[tokenizer.sep_token]
processed_inputs = input_preprocessor.preprocess(prompt)
assert sep_token_id in processed_inputs["prompt_token_ids"]