Abstract
Techniques are presented herein that support the automatic generation of refined and summarized text from a system logging (syslog) message sequence. Aspects of the presented techniques employ an abstractive syslog summarization large language model (LLM) that is trained with contrastive learning and then fine-tuned using a Low-Rank Adaptation (LoRA) methodology. Under further aspects of the presented techniques, auxiliary text (such as network incident reports and application incident reports) is added to the prompt of the input of the LLM model to help the model generate a richer syslog summarization.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Jiao, Cheng; Yang, Bruce; Chen, Lynn; Wang, Xinqi; and Shao, Qihong, "NETWORK DEVICE SYSTEM LOGGING SUMMARIZATION BASED ON LOW-RANK ADAPTATION AND CONTRASTIVE LEARNING", Technical Disclosure Commons, (August 11, 2023)
https://www.tdcommons.org/dpubs_series/6130