LECAT:从预训练数据衡量 LLM 的类别知识学习
LECAT: Measuring LLM Category Learning from Pretraining Data
AI 导读
研究者提出 LeCat 框架,从记忆域内类别知识、泛化到未见实例、灵活运用知识三个维度评估 LLM 的类别学习能力,并追踪其如何从预训练文本的各类“知识表达”形式中习得。
正文 · 原文
Abstract:Humans know that airplanes are fast, that this extends to red airplanes, but that it does not hold for broken airplanes. How do LLMs acquire category knowledge that humans learn and apply effortlessly? We introduce LeCat, a two-part framework to analyze LLM's Learning of Category knowledge. First, we use LeCat to evaluate models on their ability to learn category knowledge, distinguishing between three cases: memorization of in-domain category knowledge, generalization to unseen instances, and ability to flexibly use knowledge in appropriate contexts. Then, we use LeCat to trace how this knowledge can be learned in pretraining from text that may encode category knowledge in various ways, such as verbatim repetition or exceptions, which we refer to as forms of "knowledge expression". We first assess open-source pretrained models using 199 curated category-feature pairs and analyze their training data. Our findings suggest that strengthening context and category-feature co-occurrence likelihood (PMI) improve learning for larger models and verbatim rule-like statements diminish flexible category learning. To systematically isolate these effects, we then train a 68M model from scratch with precise control over the categorical information each sees in their training data, confirming that verbatim repetition reduces flexible learning but mixture of diverse knowledge expressions lead to more robust category learning. Together, our findings highlight that while contextual flexibility is persistently difficult for models, how category knowledge is expressed in pretraining greatly impacts models' acquisition of robust category knowledge.
| Comments: | 21 pages, 4 figures, 8 tables |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.18419 [cs.AI] |
| (or arXiv:2611.18419v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.18419 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Jimin Mun [view email]
[v1]
Fri, 13 Nov 2026 16:59:46 UTC (867 KB)
来源:arXiv cs.AI · arxiv.org