[CI/Build] Split out mm processor tests (#23260)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>

.buildkite/test-pipeline.yaml

@@ -545,6 +545,15 @@ steps:
   commands:
     - pytest -v -s models/language/pooling -m 'not core_model'
 
+- label: Multi-Modal Processor Test
+  source_file_dependencies:
+  - vllm/
+  - tests/models/multimodal
+  commands:
+    - pip install git+https://github.com/TIGER-AI-Lab/Mantis.git
+    - pytest -v -s models/multimodal/processing --ignore models/multimodal/processing/test_tensor_schema.py
+    - pytest -v -s models/multimodal/processing/test_tensor_schema.py
+
 - label: Multi-Modal Models Test (Standard)
   mirror_hardwares: [amdexperimental]
   torch_nightly: true
@@ -554,9 +563,7 @@ steps:
   commands:
     - pip install git+https://github.com/TIGER-AI-Lab/Mantis.git
     - pip freeze | grep -E 'torch'
-    - pytest -v -s models/multimodal/processing
-    - pytest -v -s --ignore models/multimodal/generation/test_whisper.py --ignore models/multimodal/test_tensor_schema.py models/multimodal -m core_model
-    - pytest -v -s models/multimodal/test_tensor_schema.py -m core_model  # Needs mp_method="spawn"
+    - pytest -v -s models/multimodal -m core_model --ignore models/multimodal/generation/test_whisper.py --ignore models/multimodal/processing
     - cd .. && pytest -v -s tests/models/multimodal/generation/test_whisper.py -m core_model  # Otherwise, mp_method="spawn" doesn't work
 
 - label: Multi-Modal Models Test (Extended) 1
@@ -567,7 +574,7 @@ steps:
   - tests/models/multimodal
   commands:
     - pip install git+https://github.com/TIGER-AI-Lab/Mantis.git
-    - pytest -v -s --ignore models/multimodal/generation/test_common.py --ignore models/multimodal/processing models/multimodal -m 'not core_model'
+    - pytest -v -s models/multimodal -m 'not core_model' --ignore models/multimodal/generation/test_common.py --ignore models/multimodal/processing
 
 - label: Multi-Modal Models Test (Extended) 2
   mirror_hardwares: [amdexperimental]

tests/models/multimodal/test_tensor_schema.py → tests/models/multimodal/processing/test_tensor_schema.py

@@ -24,9 +24,9 @@ from vllm.utils import GiB_bytes, is_list_of, set_default_torch_num_threads
 from vllm.v1.core.kv_cache_utils import get_kv_cache_config
 from vllm.v1.engine.core import EngineCore as V1EngineCore
 
-from ...conftest import VllmRunner
-from ..registry import _MULTIMODAL_EXAMPLE_MODELS, HF_EXAMPLE_MODELS
-from ..utils import dummy_hf_overrides
+from ....conftest import VllmRunner
+from ...registry import _MULTIMODAL_EXAMPLE_MODELS, HF_EXAMPLE_MODELS
+from ...utils import dummy_hf_overrides
 
 ARCH_TO_SKIP = {
     "MolmoForCausalLM": "incompatible requirements",
@@ -147,7 +147,6 @@ def get_model_id_to_test(
     return filtered_results
 
 
-@pytest.mark.core_model
 @pytest.mark.parametrize(
     "model_arch, model_id",
     get_model_id_to_test(_MULTIMODAL_EXAMPLE_MODELS.keys()))

vllm/model_executor/models/cohere2_vision.py

@@ -170,6 +170,8 @@ class Cohere2VisionProcessingInfo(BaseProcessingInfo):
 
         # The current implementation of get_number_of_image_patches
         # is incorrect, so we patch it here.
+        # TODO: Revert once
+        # https://github.com/huggingface/transformers/pull/40312 is released.
         # return image_processor.get_number_of_image_patches(image_height,
         #                                                    image_width, {})