[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 -7
View File
@@ -19,7 +19,7 @@ from typing_extensions import TypedDict
from vllm.compilation.decorators import support_torch_compile
from vllm.config import VllmConfig
from vllm.config.multimodal import AudioDummyOptions, BaseDummyOptions
from vllm.config.multimodal import MultiModalDummyOptions
from vllm.distributed import (
get_pp_group,
get_tensor_model_parallel_world_size,
@@ -1341,16 +1341,13 @@ class MossAudioDummyInputsBuilder(BaseDummyInputsBuilder[MossAudioProcessingInfo
self,
seq_len: int,
mm_counts: Mapping[str, int],
mm_options: Mapping[str, BaseDummyOptions],
mm_options: MultiModalDummyOptions,
) -> MultiModalDataDict:
num_audios = mm_counts.get("audio", 0)
audio_overrides = mm_options.get("audio")
assert audio_overrides is None or isinstance(audio_overrides, AudioDummyOptions)
return {
"audio": self._get_dummy_audios(
length=16000,
num_audios=num_audios,
overrides=audio_overrides,
num_audios=mm_counts.get("audio", 0),
overrides=mm_options.get("audio"),
)
}