Large Language Models (LLMs) often retain inaccurate or outdated information from pre-training, leading to incorrect predictions or biased outputs during inference. While existing model editing methods can address this challenge, they struggle with editing large amounts of factual information simultaneously and may compromise the general capabilities of the models. In this paper, our empirical study demonstrates that it is feasible to edit the internal representations of LLMs and replace the entities in a manner similar to editing natural language inputs. Based on this insight, we introduce the Latent Knowledge Scalpel (LKS), an LLM editor that manipulates the latent knowledge of specific entities via a lightweight hypernetwork to enable precise and large-scale editing. Experiments conducted on Llama-2 and Mistral show even with the number of simultaneous edits reaching 10,000, LKS effectively performs knowledge editing while preserving the general abilities of the edited LLMs. Code is available at: https://github.com/Linuxin-xxx/LKS.

Microservices architecture has gained popularity in modern software development due to its scalability and flexibility. However, understanding the complexity of interactions and dependencies between services presents significant challenges, which complicates the identification and analysis of errors within microservice applications. To gain insights into the architecture and interdependencies of microservices applications, prior studies have developed dependency graphs to illustrate the relationships among services. However, the methods used to construct these dependency graphs are not suitable for common microservices applications and suffer from insufficient data granularity. To address these shortcomings, we introduce LLM4MDG, an in-novative framework for constructing microservices dependency graphs using an LLM-driven multi-agent system. By leveraging optimized prompt engineering and principles of knowledge graphs, LLM4MDG can effectively identify and interpret service interactions across diverse microservice ecosystems, achieving high accuracy and adaptability across various scenarios. We also present a new open-source dataset comprising 47 microservices applications, annotated by domain experts, to validate our frame-work. Evaluation results demonstrate that LLM4MDG achieves an 88.3% accuracy in identifying data dependencies in the Train Ticket project, a benchmark application with over 80 service instances. This study provides a robust solution for constructing dependency graphs and facilitating better system understanding and management.

Despite the remarkable progress made in synthesizing emotional speech from text, it is still challenging to provide emotion information to existing speech segments. Previous methods mainly rely on parallel data, and few works have studied the generalization ability for one model to transfer emotion information across different languages. To cope with such problems, we propose an emotion transfer system named ET-GAN, for learning language-independent emotion transfer from one emotion to another without parallel training samples. Based on cycle-consistent generative adversarial network, our method ensures the transfer of only emotion information across speeches with simple loss designs. Besides, we introduce an approach for migrating emotion information across different languages by using transfer learning. The experiment results show that our method can efficiently generate high-quality emotional speech for any given emotion category, without aligned speech pairs.