Adaptive search improves hardware-constrained large language model inference

ACSeL: Adaptive Configuration Search for Hardware-Constrained LLM Inference

Machine Learning

Summary

Running large language models on devices with limited memory and speed is tough because you need to balance many settings like model size and precision. The authors developed ACSeL, a method that quickly finds the best settings to meet memory and speed limits while keeping output quality high. They introduced a way to predict quality using a small sample, making the search faster. Their technique works well on both powerful and edge hardware, helping ensure models run efficiently without too much trial and error.

What this means in practice

  • For edge device engineers: Configure large language models to best fit memory and latency limits on edge hardware while preserving output quality.
  • For cloud infrastructure teams: Efficiently select model and quantization settings to satisfy strict service-level objectives without exhaustive testing.

Authors

Bokun Wang, Hung-Yueh Chiang, Diana Marculescu

Abstract

Serving large language models (LLMs) under tight memory and latency service-level objectives (SLOs), including on edge devices, requires jointly selecting the model size, batch size, quantization algorithm, and mixed-precision quantization recipes. Exhaustive search over this combinatorially large space is infeasible. Unlike data-center serving, which often prioritizes throughput or generation speed, resource-constrained deployments typically set memory and latency budgets as constraints and then seek the configuration with the highest generation quality among feasible ones. To this end, we propose ACSeL, a framework that identifies configurations predicted to satisfy deployment constraints while achieving high predicted generation quality. To enable efficient quality estimation, we use Relative Model Quality Degradation (RMQD), a proxy based on the divergence between the candidate and reference models' next-token distributions over a small calibration corpus. Across the evaluated model sizes and precisions, RMQD is inexpensive to evaluate and strongly correlated with downstream benchmark performance. ACSeL combines (i) a lightweight quality surrogate that predicts RMQD directly from a configuration; (ii) Few-shot Median Calibration (FMC), which calibrates operator-profile-based latency estimates using a small number of end-to-end profiles; and (iii) a structured search algorithm BandPareto for exploring layerwise mixed-precision configurations. Experiments on an NVIDIA RTX A6000 workstation and an NVIDIA Jetson Thor edge platform show that FMC substantially improves deployment feasibility, while BandPareto + FMC achieves the highest feasibility and the lowest RMQD gaps to the best measured configurations in the evaluation set.