HADIS: Hybrid Adaptive Diffusion Model Serving for Efficient Text-to-Image Generation

2025-08-30

AI Summary

Text-to-image diffusion models incur high inference costs, hindering real-time deployment. Existing cascaded serving systems employ static configurations, mandating all queries to pass through a lightweight model for initial screening—wasting GPU resources on semantically complex prompts. This work proposes a hybrid adaptive serving system: (1) a rule-driven prompt router that directly routes semantically complex queries to heavyweight models; (2) an offline performance analysis to construct a Pareto-optimal configuration table, enabling runtime-dynamic cascade decisions; and (3) joint optimization of query routing, model selection, and GPU resource allocation. Experiments under realistic workloads demonstrate that, compared to baseline systems, our approach improves response quality by up to 35% and reduces latency violation rates by 2.7×–45×.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Planning with Language Models
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web search models and ranking
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

Text-to-image diffusion models have achieved remarkable visual quality but incur high computational costs, making real-time, scalable deployment challenging. Existing query-aware serving systems mitigate the cost by cascading lightweight and heavyweight models, but most rely on a fixed cascade configuration and route all prompts through an initial lightweight stage, wasting resources on complex queries. We present HADIS, a hybrid adaptive diffusion model serving system that jointly optimizes cascade model selection, query routing, and resource allocation. HADIS employs a rule-based prompt router to send clearly hard queries directly to heavyweight models, bypassing the overhead of the lightweight stage. To reduce the complexity of resource management, HADIS uses an offline profiling phase to produce a Pareto-optimal cascade configuration table. At runtime, HADIS selects the best cascade configuration and GPU allocation given latency and workload constraints. Empirical evaluations on real-world traces demonstrate that HADIS improves response quality by up to 35% while reducing latency violation rates by 2.7-45$ imes$ compared to state-of-the-art model serving systems.

Problem

• Optimizes cascade model selection for text-to-image generation
• Reduces computational costs in diffusion model serving
• Improves resource allocation and query routing efficiency

Innovation

• Hybrid adaptive diffusion model serving system
• Rule-based prompt router for query routing
• Offline profiling for Pareto-optimal cascade configuration