[Multimodal] Type dummy options per modality (#57576)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Harry Mellor
2026-09-18 17:29:11 +01:00
committed by GitHub
co-authored by Claude Opus 5
parent 56cac80b14
commit 8a5cf54387
93 changed files with 461 additions and 720 deletions
+4 -8
View File
@@ -15,7 +15,7 @@ from transformers import (
)
from vllm.config import CacheConfig, VllmConfig
from vllm.config.multimodal import BaseDummyOptions, ImageDummyOptions
from vllm.config.multimodal import MultiModalDummyOptions
from vllm.inputs import MultiModalDataDict
from vllm.model_executor.layers.activation import get_act_fn
from vllm.model_executor.layers.quantization import QuantizationConfig
@@ -444,23 +444,19 @@ class Blip2DummyInputsBuilder(BaseDummyInputsBuilder[Blip2ProcessingInfo]):
self,
seq_len: int,
mm_counts: Mapping[str, int],
mm_options: Mapping[str, BaseDummyOptions],
mm_options: MultiModalDummyOptions,
) -> MultiModalDataDict:
hf_config = self.info.get_hf_config()
vision_config = hf_config.vision_config
max_image_size = vision_config.image_size
num_images = mm_counts.get("image", 0)
image_overrides = mm_options.get("image")
assert image_overrides is None or isinstance(image_overrides, ImageDummyOptions)
return {
"image": self._get_dummy_images(
width=max_image_size,
height=max_image_size,
num_images=num_images,
overrides=image_overrides,
num_images=mm_counts.get("image", 0),
overrides=mm_options.get("image"),
)
}