MASSV: Multimodal Adaptation and Self-Data Distillation for Speculative Decoding of Vision-Language Models
This addresses the problem of slow inference in VLMs for applications requiring real-time multimodal processing, though it is incremental as it adapts existing speculative decoding techniques to a new domain.
The paper tackled the challenge of applying speculative decoding to vision-language models (VLMs) by introducing MASSV, which transforms small language models into multimodal drafters, resulting in up to 30% increased accepted length and 1.46x inference speedups on visually-grounded tasks.
Speculative decoding significantly accelerates language model inference by enabling a lightweight draft model to propose multiple tokens that a larger target model verifies simultaneously. However, applying this technique to vision-language models (VLMs) presents two fundamental challenges: small language models that could serve as efficient drafters lack the architectural components to process visual inputs, and their token predictions fail to match those of VLM target models that consider visual context. We introduce Multimodal Adaptation and Self-Data Distillation for Speculative Decoding of Vision-Language Models (MASSV), which transforms existing small language models into effective multimodal drafters through a two-phase approach. MASSV first connects the target VLM's vision encoder to the draft model via a lightweight trainable projector, then applies self-distilled visual instruction tuning using responses generated by the target VLM to align token predictions. Comprehensive experiments across the Qwen2.5-VL and Gemma3 model families demonstrate that MASSV increases accepted length by up to 30% and delivers end-to-end inference speedups of up to 1.46x on visually-grounded tasks. MASSV provides a scalable, architecture-compatible method for accelerating both current and future VLMs.