Learning to Inference Adaptively for Multimodal Large Language Models
This work addresses efficiency for deploying MLLMs in resource-constrained environments, representing an incremental improvement by adapting to runtime conditions.
The paper tackles the high computational cost of Multimodal Large Language Models (MLLMs) in resource-constrained settings by introducing AdaLLaVA, an adaptive inference framework that dynamically reconfigures operations based on input data and latency budgets, achieving varying accuracy-latency tradeoffs in benchmarks like question-answering and reasoning.
Multimodal Large Language Models (MLLMs) have shown impressive capabilities in visual reasoning, yet come with substantial computational cost, limiting their deployment in resource-constrained settings. Despite recent effort on improving the efficiency of MLLMs, prior solutions fall short in responding to varying runtime conditions, in particular changing resource availability (e.g., contention due to the execution of other programs on the device). To bridge this gap, we introduce AdaLLaVA, an adaptive inference framework that learns to dynamically reconfigure operations in an MLLM during inference, accounting for the input data and a latency budget. We conduct extensive experiments across benchmarks involving question-answering, reasoning, and hallucination. Our results show that AdaLLaVA effectively adheres to input latency budget, achieving varying accuracy and latency tradeoffs at runtime. Further, we demonstrate that AdaLLaVA adapts to both input latency and content, can be integrated with token selection for enhanced efficiency, and generalizes across MLLMs. Our project webpage with code release is at https://zhuoyan-xu.github.io/ada-llava/.