Automatic Configuration of LLM Post-Training Pipelines

Abstract

LLM post-training pipelines that combine supervised fine-tuning and reinforcement learning are difficult to configure under realistic compute budgets: the configuration space is high-dimensional and heterogeneous, stages are strongly coupled, and each end-to-end evaluation is expensive. We propose AutoPipe, a budget-aware two-stage framework for configuration selection in LLM post-training. Offline, AutoPipe learns a dataset-conditioned learning-to-rank surrogate from historical runs, capturing within-dataset preferences and providing transferable guidance toward promising regions of the configuration space. Online, for a new dataset, AutoPipe uses the offline guidance to steer Bayesian optimization and models dataset-specific deviations with a Gaussian-process residual surrogate. To reduce evaluation cost, each trial is early-stopped and scored by a learned predictor that maps early training signals to a low-cost proxy for final post-training performance. Experiments on biomedical reasoning tasks show that AutoPipe consistently outperforms offline-only baselines and achieves comparable performance with the strongest online HPO baselines while using less than 10\% of their computational cost.

Figures (5)

• Figure 1: Overview of the proposed two-phase configuration selection framework.
• Figure 2: Sensitivity of performance (average benchmark accuracy) to online evaluation budget.
• Figure C.1: Prompt template used for LLM Recommendation (Description-only) baseline. Values enclosed in angled brackets are dependent on the unseen dataset.
• Figure C.2: The full history-conditioned prompt template, utilizing meta-feature-based retrieval to provide historical experiment results (nearest neighbors) to the LLM surrogate.
• Figure D.1: Sensitivity of configuration selection performance to the online evaluation budget across ten held-out datasets.