paper-with-me

홈 › Papers

How do language models learn facts? Dynamics, curricula and hallucinations

2025-03-27 · Nicolas Zucchet, Jörg Bornschein, Stephanie Chan, Andrew Lampinen, Razvan Pascanu, Soham De

Large language models accumulate vast knowledge during pre-training, yet the dynamics governing this acquisition remain poorly understood. This work investigates the learning dynamics of language models on a synthetic factual recall task, uncovering three key findings: First, language models learn in three phases, exhibiting a performance plateau before acquiring precise factual knowledge. Mechanistically, this plateau coincides with the formation of attention-based circuits that support recall. Second, the training data distribution significantly impacts learning dynamics, as imbalanced distributions lead to shorter plateaus. Finally, hallucinations emerge simultaneously with knowledge, and integrating new knowledge into the model through fine-tuning is challenging, as it quickly corrupts its existing parametric memories. Our results emphasize the importance of data distribution in knowledge acquisition and suggest novel data scheduling strategies to accelerate neural network training.

📄 PDF Abstract BibTeX arXiv:2503.21676

Code (0)

등록된 구현이 없습니다.

Tasks

Scheduling

Similar Papers 제목 키워드 기반

On Large Language Models' Hallucination with Regard to Known Facts

2024-03-29 · Che Jiang, Biqing Qi, Xiangyu Hong, Dayuan Fu 외

Large language models are successful in answering factoid questions but are also prone to hallucination. We investigate the phenomenon of LLMs possessing correct answer knowledge yet still hallucinating from the perspect…

HallucinationTriplet

Calibrated Language Models Must Hallucinate

2023-11-24 · Adam Tauman Kalai, Santosh S. Vempala

Recent language models generate false but plausible-sounding text with surprising frequency. Such "hallucinations" are an obstacle to the usability of language-based AI systems and can harm people who rely upon their out…

ArticlesHallucination

Don't Fight Hallucinations, Use Them: Estimating Image Realism using NLI over Atomic Facts

2025-03-20 · Elisei Rykov, Kseniia Petrushina, Kseniia Titova, Alexander Panchenko 외

Quantifying the realism of images remains a challenging problem in the field of artificial intelligence. For example, an image of Albert Einstein holding a smartphone violates common-sense because modern smartphone were …

Common Sense ReasoningNatural Language Inference

On Early Detection of Hallucinations in Factual Question Answering

2023-12-19 · Ben Snyder, Marius Moisescu, Muhammad Bilal Zafar

While large language models (LLMs) have taken great strides towards helping humans with a plethora of tasks, hallucinations remain a major impediment towards gaining user trust. The fluency and coherence of model generat…

HallucinationOpen-Ended Question AnsweringQuestion Answering

AI Hallucination from Students' Perspective: A Thematic Analysis

2026-01-11 · Abdulhadi Shoufan, Ahmad-Azmi-Abdelhamid Esmaeil arxiv

As students increasingly rely on large language models, hallucinations pose a growing threat to learning. To mitigate this, AI literacy must expand beyond prompt engineering to address how students should detect and resp…

Prompt Engineering