An open-source large language model for Chinese education research
Peer-Reviewed Publication
Updates every hour. Last Updated: 6-Dec-2025 04:11 ET (6-Dec-2025 09:11 GMT/UTC)
Open-source large language models (LLMs) research has made significant progress, but most studies predominantly focus on general-purpose English data, which poses challenges for LLM research in Chinese education. To address this, this research first reviewed and synthesized the core technologies of representative open-source LLMs, and designed an advanced 1.5B-parameter LLM tailored for the Chinese education field. Chinese education large language model (CELLM) is trained from scratch, involving two stages, namely, pre-training and instruction fine-tuning. In the pre-training phase, an open-source dataset is utilized for the Chinese education domain. During the instruction fine-tuning stage, the Chinese instruction dataset is developed and open-sourced, comprising over 258,000 data entries. Finally, the results and analysis of CELLM across multiple evaluation datasets are presented, which provides a reference baseline performance for future research. All of the models, data, and codes are open-source to foster community research on LLMs in the Chinese education domain.
Nature is a source of well-being and recovery for many people. However, research shows that there is also a growing number of individuals who experience negative emotions, such as fear, discomfort, or even disgust, toward nature. The phenomenon, called biophobia, is now highlighted in a new study from Lund University.
The Quantum Education Summit event, hosted at CosmoCaixa Science Museum of Barcelona, will gather more than 300 participants from five different continents.
The topics of the conference will cover the full spectrum of quantum education: from foundational curriculum design, workforce training and inclusive global participation to policy strategies, educational innovations, and outreach efforts that advance accessibility and broaden the impact of quantum learning worldwide.
Keynote speakers include Nobel Laureate Carl Wieman, from Stanford University, Maité Depuis, from Perimeter Institute of Canada, and Eboney Hearn, from MIT Introduction to Technology, Engineering and Science.
Glycolipid metabolic disorders, linked to cardiovascular diseases and cancer, are a major global health challenge. Current single-disease treatments remain unsatisfied in reducing long-term risks. In 2024, Professor Jiao Guo along with global experts launched the "Global Initiative for Glycolipid Metabolic Health" to enhance prevention through scientific research, public education, and integrated management systems.
A new study suggests that traditional learning activities like making notes remain critical for students’ reading comprehension and retention, while also suggesting that large language models (LLMs), such as ChatGPT or Microsoft Copilot, could be a useful tool for helping students clarify, explore, and contextualise learning material.
A new study from researchers at Boston University Chobanian & Avedisian School of Medicine has found that while some areas of forensic anthropology appear to have tacit agreement on standardized methods and citations, there are many areas where forensic anthropologists are divided and lack standardization. The researchers believe these areas of division and disagreement-for example, thoughts on the objectivity of practitioners and education and training requirements, demonstrate the need for a robust and enforceable standardization that applies to all forensic anthropologists.