CoServe: Efficient Collaboration-of-Experts (CoE) Model Inference with Limited Memory
This work addresses efficiency problems for deploying specialized AI models in memory-constrained environments like manufacturing, but it is incremental as it builds on existing CoE approaches.
The paper tackles the memory and performance challenges of deploying Collaboration-of-Experts (CoE) model serving systems by proposing CoServe, which reduces expert switching and optimizes resource allocation, achieving 4.5x to 12x higher throughput in real-world intelligent manufacturing workloads.
Large language models like GPT-4 are resource-intensive, but recent advancements suggest that smaller, specialized experts can outperform the monolithic models on specific tasks. The Collaboration-of-Experts (CoE) approach integrates multiple expert models, improving the accuracy of generated results and offering great potential for precision-critical applications, such as automatic circuit board quality inspection. However, deploying CoE serving systems presents challenges to memory capacity due to the large number of experts required, which can lead to significant performance overhead from frequent expert switching across different memory and storage tiers. We propose CoServe, an efficient CoE model serving system on heterogeneous CPU and GPU with limited memory. CoServe reduces unnecessary expert switching by leveraging expert dependency, a key property of CoE inference. CoServe introduces a dependency-aware request scheduler and dependency-aware expert management for efficient inference. It also introduces an offline profiler to automatically find optimal resource allocation on various processors and devices. In real-world intelligent manufacturing workloads, CoServe achieves 4.5$\times$ to 12$\times$ higher throughput compared to state-of-the-art systems.