Two-dimensional(2D)van der Waals magnetic materials have promising and versatile electronic and magnetic properties in the 2D limit,indicating a considerable potential to advance spintronic applications.Theoretical pr...Two-dimensional(2D)van der Waals magnetic materials have promising and versatile electronic and magnetic properties in the 2D limit,indicating a considerable potential to advance spintronic applications.Theoretical predictions thus far have not ascertained whether monolayer VCl_(3) is a ferromagnetic(FM)or anti-FM monolayer;this also remains to be experimentally verified.We theoretically investigate the influence of potential factors,including C_(3) symmetry breaking,orbital ordering,epitaxial strain,and charge doping,on the magnetic ground state.Utilizing first-principles calculations,we predict a collinear type-Ⅲ FM ground state in monolayer VCl_(3) with a broken C_(3) symmetry,wherein only the former two of three t_(2g)orbitals(a_(1g),e_(g2)^(π)and e_(g1)^(π))are occupied.The atomic layer thickness and bond angles of monolayer VCl_(3) undergo abrupt changes driven by an orbital ordering switch,resulting in concomitant structural and magnetic phase transitions.Introducing doping to the underlying Cl atoms of monolayer VCl_(3) without C_(3) symmetry simultaneously induces in-and out-of-plane polarizations.This can achieve a multiferroic phase transition if combined with the discovered adjustments of magnetic ground state and polarization magnitude under strain.The establishment of an orbital-ordering driven regulatory mechanism can facilitate deeper exploration and comprehension of magnetic properties of strongly correlated systems in monolayer VCl_(3).展开更多
预训练语言模型(pre-trained languages model,PTLM)在自然语言处理(natural language processing,NLP)领域取得了令人瞩目的成功,并由此引发了下游任务从监督学习到预训练-微调范式的转变。在此之后,一系列预训练模型的创新研究涌现出...预训练语言模型(pre-trained languages model,PTLM)在自然语言处理(natural language processing,NLP)领域取得了令人瞩目的成功,并由此引发了下游任务从监督学习到预训练-微调范式的转变。在此之后,一系列预训练模型的创新研究涌现出来。本文系统性、全面的回顾了自然语言处理的代表性工作和最新进展,并按照类别系统性的介绍了自然语言处理领域的预训练模型。首先我们简要介绍了预训练模型,以及不同的模型特点和框架。之后,我们介绍并分析了预训练模型的影响和挑战以及下游任务中的应用。最后,我们简要总结并阐述了预训练模型未来的研究方向。展开更多
基金This work was supported by the National Natural Science Foundation of China(Grant No.11802242)the Fundamental Research Funds for the Central Universities(Grant No.D5000230066).
基金supported by the National Key Research and Development Program of China(Grant Nos.2018YFE0202700 and 2023YFA1406500)the National Natural Science Foundation of China(Grant Nos.11974422 and 12104504)+2 种基金the Strategic Priority Research Program of the Chinese Academy of Sciences(Grant No.XDB30000000)Fundamental Research Funds for the Central Universities,and Research Funds of Renmin University,China(Grant No.22XNKJ30)supported by the Outstanding Innovative Talents Cultivation Funded Programs 2023 of Renmin University,China。
文摘Two-dimensional(2D)van der Waals magnetic materials have promising and versatile electronic and magnetic properties in the 2D limit,indicating a considerable potential to advance spintronic applications.Theoretical predictions thus far have not ascertained whether monolayer VCl_(3) is a ferromagnetic(FM)or anti-FM monolayer;this also remains to be experimentally verified.We theoretically investigate the influence of potential factors,including C_(3) symmetry breaking,orbital ordering,epitaxial strain,and charge doping,on the magnetic ground state.Utilizing first-principles calculations,we predict a collinear type-Ⅲ FM ground state in monolayer VCl_(3) with a broken C_(3) symmetry,wherein only the former two of three t_(2g)orbitals(a_(1g),e_(g2)^(π)and e_(g1)^(π))are occupied.The atomic layer thickness and bond angles of monolayer VCl_(3) undergo abrupt changes driven by an orbital ordering switch,resulting in concomitant structural and magnetic phase transitions.Introducing doping to the underlying Cl atoms of monolayer VCl_(3) without C_(3) symmetry simultaneously induces in-and out-of-plane polarizations.This can achieve a multiferroic phase transition if combined with the discovered adjustments of magnetic ground state and polarization magnitude under strain.The establishment of an orbital-ordering driven regulatory mechanism can facilitate deeper exploration and comprehension of magnetic properties of strongly correlated systems in monolayer VCl_(3).
文摘预训练语言模型(pre-trained languages model,PTLM)在自然语言处理(natural language processing,NLP)领域取得了令人瞩目的成功,并由此引发了下游任务从监督学习到预训练-微调范式的转变。在此之后,一系列预训练模型的创新研究涌现出来。本文系统性、全面的回顾了自然语言处理的代表性工作和最新进展,并按照类别系统性的介绍了自然语言处理领域的预训练模型。首先我们简要介绍了预训练模型,以及不同的模型特点和框架。之后,我们介绍并分析了预训练模型的影响和挑战以及下游任务中的应用。最后,我们简要总结并阐述了预训练模型未来的研究方向。
