Browsing by Author "Sun S"

Now showing 1 - 2 of 2

A Framework to Assess Multilingual Vulnerabilities of LLMs
(Association for Computing Machinery, 2025-05-23) Tang L; Bogahawatta N; Ginige Y; Xu J; Sun S; Ranathunga S; Seneviratne S
Large Language Models (LLMs) are acquiring a wider range of capabilities, including understanding and responding in multiple languages. While they undergo safety training to prevent them from answering illegal questions, imbalances in training data and human evaluation resources can make these models more susceptible to attacks in low-resource languages (LRL). This paper proposes a framework to automatically assess the multilingual vulnerabilities of commonly used LLMs. Using our framework, we evaluated six LLMs across eight languages representing varying levels of resource availability. We validated the assessments generated by our automated framework through human evaluation in two languages, demonstrating that the framework's results align with human judgments in most cases. Our findings reveal vulnerabilities in LRL; however, these may pose minimal risk as they often stem from the model's poor performance, resulting in incoherent responses.
Integrated transcriptome and proteome analyses reveal potential mechanisms in Stipa breviflora underlying adaptation to grazing
(John Wiley and Sons Australia, Ltd on behalf of Chinese Grassland Society and Lanzhou University, 2024-03-14) Liu Y; Sun S; Zhang Y; Song M; Tian Y; Lockhart PJ; Zhang X; Xu Y; Dang Z; Matthew C
Background: Long-term overgrazing has led to severe degradation of grasslands, posing a significant threat to the sustainable use of grassland resources. Methods: Based on the investigation of changes in functional traits and photosynthetic physiology of Stipa breviflora under no grazing, moderate grazing, and heavy grazing treatments, the changes in expression patterns of genes and proteins associated with different grazing intensities were assessed through integrative transcriptomic and proteomic analyses. Results: Differentially expressed genes and proteins were identified under different grazing intensities. They were mainly related to RNA processing, carbon metabolism, and secondary metabolite biosynthesis. These findings suggest that long-term grazing leads to molecular phenotypic plasticity, affecting various biological processes and metabolic pathways in S. breviflora. Correlation analysis revealed low correlation between the transcriptome and the proteome, indicating a large-scale regulation of gene expression at the posttranscriptional and translational levels during the response of S. breviflora to grazing. The expression profiles of key genes and proteins involved in photosynthesis and phenylpropanoid metabolism pathways suggested their synergistic response to grazing in S. breviflora. Conclusions: Our study provides insight into the adaptation mechanisms of S. breviflora to grazing and provides a scientific basis for the development of more efficient grassland protection and utilization practices.