{"slug": "littlelearner-language-models-under-pedagogically-controlled-knowledge-exposure", "title": "LittleLearner: Language Models Under Pedagogically Controlled Knowledge Exposure", "summary": "Researchers released LittleCurriculum, an 88B-token pretraining corpus restricted to U.S. elementary school material and explicitly excluding concepts, facts and vocabulary taught above Grade 5, alongside LittleLearner, a 5B-parameter LLM trained from scratch on it. The team reports that post-training and in-context learning let LittleLearner better use existing knowledge but did not raise out-of-scope capabilities, and released both the corpus and model as a developmentally restricted sandbox for studying how models acquire, represent and use data under a defined training scope.", "body_md": "arXiv:2608.13545v3 Announce Type: replace \nAbstract: Modern language models are trained on heterogeneous web-scale text corpora. Consequently, studying knowledge and skill acquisition is difficult, as prior exposure to related content is hard to characterize. To address this challenge, we introduce LittleCurriculum, a curated 88B-token pretraining corpus tailored to U.S. elementary school material, explicitly excluding concepts, facts, and vocabulary taught above Grade 5. Training a 5B-parameter LLM from scratch on LittleCurriculum yields LittleLearner, a model with sufficient language competence for open-ended evaluation, yet with clear knowledge and capability boundaries mapped to interpretable curriculum guidelines. We release LittleCurriculum and LittleLearner as a developmentally restricted sandbox to study how models acquire, represent, and use data under a well-defined training scope. We illustrate the sandbox's utility in a first suite of experiments on injecting new knowledge through post-training and in-context learning. These methods let LittleLearner better utilize existing knowledge, but do not raise out-of-scope capabilities. Our findings underscore the value of this controlled environment for future investigations.", "url": "https://wpnews.pro/news/littlelearner-language-models-under-pedagogically-controlled-knowledge-exposure", "canonical_source": "https://www.machinebrief.com/news/littlelearner-language-models-under-pedagogically-controlled-qzmb", "published_at": "2026-11-09 05:00:00+00:00", "updated_at": "2026-11-09 05:23:12.741866+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "large-language-models", "ai-research"], "entities": ["LittleCurriculum", "LittleLearner", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/littlelearner-language-models-under-pedagogically-controlled-knowledge-exposure", "markdown": "https://wpnews.pro/news/littlelearner-language-models-under-pedagogically-controlled-knowledge-exposure.md", "text": "https://wpnews.pro/news/littlelearner-language-models-under-pedagogically-controlled-knowledge-exposure.txt", "jsonld": "https://wpnews.pro/news/littlelearner-language-models-under-pedagogically-controlled-knowledge-exposure.jsonld"}}