Does sglang support deploying multimodal models with spec decoding feature?

[HPUedCSLearner]

I am a newbie to sglang.

I would like to ask you, when deploying a multimodal model using sglang, such as qwen2.5vl, can I use the features of Eagle speculative inference?

If possible, please give me some guidance, thank you very much!!!