Search papers, labs, and topics across Lattice.
This survey systematically catalogs 80 methods for Unsupervised Post-Training (UPT) of foundation models, focusing on adaptation using unlabeled inputs and internal model artifacts rather than external supervision. The authors highlight how the choice of internal signal and task structure can either enhance model performance or exacerbate errors, providing critical insights into the risks and benefits of UPT. Additionally, they introduce a unified framework that maps deployment regimes through the lens of Input Visibility and Update Persistence, facilitating better selection and evaluation of UPT strategies.
UPT can either refine model capabilities or amplify errors, depending on the internal signals used during adaptation.
Foundation-model post-training usually relies on human labels, preference data, stronger teachers, or executable verifiers. We study Unsupervised Post-Training (UPT): update-bearing adaptation on unlabeled inputs whose learning signal is derived from same-lineage model artifacts rather than an external oracle. We catalog 80 strict UPT methods and organize them by the object that supplies the update signal: a prediction statistic, a sample relation, a self-generated target, or an internal evaluator. Beyond inventory, we show how the choice of internal signal and task structure determines whether post-training improves the model or recursively amplifies error. An orthogonal Input Visibility $\times$ Update Persistence view maps deployment regimes and defines a unified framework for UPT selection and evaluation.
Tsinghua AI