From 8fb0274415062912ec56c6629a04cbac60121fe3 Mon Sep 17 00:00:00 2001 From: Shanshan Shen <467638484@qq.com> Date: Mon, 8 Jun 2026 13:57:06 +0800 Subject: [PATCH] [MM][CG] Simplify ViT CUDA graph interfaces (#44484) Signed-off-by: shen-shanshan <467638484@qq.com> --- vllm/model_executor/models/interfaces.py | 4 ++-- vllm/model_executor/models/step3_vl.py | 6 ------ 2 files changed, 2 insertions(+), 8 deletions(-) diff --git a/vllm/model_executor/models/interfaces.py b/vllm/model_executor/models/interfaces.py index 749222b084..46be838f8a 100644 --- a/vllm/model_executor/models/interfaces.py +++ b/vllm/model_executor/models/interfaces.py @@ -1524,8 +1524,8 @@ class SupportsEncoderCudaGraph(Protocol): self, mm_kwargs: dict[str, Any], ) -> str: - """Return the modality of the inputs.""" - ... + """Return the modality of the inputs (default: image-only).""" + return "image" def get_max_frames_per_video( self, diff --git a/vllm/model_executor/models/step3_vl.py b/vllm/model_executor/models/step3_vl.py index 5a28ce3d00..35a817f009 100644 --- a/vllm/model_executor/models/step3_vl.py +++ b/vllm/model_executor/models/step3_vl.py @@ -709,12 +709,6 @@ class Step3VLForConditionalGeneration( out_hidden_size=self.config.hidden_size, ) - def get_input_modality( - self, - mm_kwargs: dict[str, Any], - ) -> str: - return "image" - def get_encoder_cudagraph_budget_range( self, vllm_config: "VllmConfig",