[Multi Modal] Configurable MM Profiling (#25631)

Signed-off-by: wwl2755 <wangwenlong2755@gmail.com>
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Wenlong Wang
2025-10-03 03:59:10 -07:00
committed by GitHub
co-authored by Harry Mellor
parent 2ed3f20dba
commit 79aa244678
60 changed files with 654 additions and 99 deletions
+6 -1
View File
@@ -25,6 +25,7 @@ from transformers import (BatchFeature, CLIPVisionConfig, PretrainedConfig,
ProcessorMixin)
from vllm.config import VllmConfig
from vllm.config.multimodal import BaseDummyOptions
from vllm.logger import init_logger
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.layers.vocab_parallel_embedding import (
@@ -356,17 +357,21 @@ class Phi3VDummyInputsBuilder(BaseDummyInputsBuilder[Phi3VProcessingInfo]):
self,
seq_len: int,
mm_counts: Mapping[str, int],
mm_options: Optional[Mapping[str, BaseDummyOptions]] = None,
) -> MultiModalDataDict:
num_images = mm_counts.get("image", 0)
target_width, target_height = \
self.info.get_image_size_with_most_features()
image_overrides = mm_options.get("image") if mm_options else None
return {
"image":
self._get_dummy_images(width=target_width,
height=target_height,
num_images=num_images)
num_images=num_images,
overrides=image_overrides)
}