The performance of Large Language Models (LLMs) on downstream tasks is fundamentally constrained by the capabilities acquired during pre-training. However, traditional benchmarks like MMLU often fail to reflect a base model's plasticity in complex open-ended scenarios, leading to inefficient model selection. We address this by introducing a new task of predicting post-training potential - forecasting a base model's performance before post-training. We propose RuDE (Rubric-based Discriminative Evaluation), a unified framework that bypasses the generation gap of base models by leveraging response discrimination. Guided by our systematic 4C Taxonomy, RuDE constructs controlled contrastive pairs across diverse domains by fine-grained rubric violations. Extensive experiments demonstrate a correlation greater than 90% with post-training performance. Crucially, validation via Reinforcement Learning (RL) confirms that RuDE effectively identifies high-potential smaller models that outperform larger counterparts, offering a compute-efficient mechanism for foundation model development.
@article{arxiv.2605.11978,
title = {On Predicting the Post-training Potential of Pre-trained LLMs},
author = {Xiaoyuan Li and Yubo Ma and Kexin Yang and Moxin Li and Keqin Bao and Wenie Wang and Fuli Feng and Dayiheng Liu},
journal= {arXiv preprint arXiv:2605.11978},
year = {2026}
}