Session: 2 for 1: Annotating Log Data for Machine Learning/Get to Know PyTorch – A cutting edge open source AI framework for Machine Learning
Annotating Log Data for Machine Learning – Julia Li
Have you noticed the lack of tools while digging through thousands of log lines for the root cause? All too common for engineers, this is a slow and painful process due to the sheer amount log data being vast and messy especially if the logs capture the output of events running in parallel. Commercial log analysis tools like Splunk make it easier to dissect the logs, but significant domain knowledge and time are required to troubleshoot and identify the issue.
In this talk, I will discuss our own journey into log root cause analysis using NLP. Specifically, we’ll discuss an end-to-end NLP-based log analysis system designed to help scale out the error triage process by automatically categorizing the type of fault in each log file.
Get to know PyTorch – a cutting edge open source AI framework for Machine Learning – Sahdev Zala
PyTorch has emerged as a leading open source framework for AI research and commercial production for key facets of machine learning. Created initially at Meta, it has moved under the Linux Foundation umbrella as the PyTorch Foundation to accelerate progress in AI. The foundation was created in September 2022, and in early 2023 the project announced a first major update with the release of PyTorch 2.0. PyTorch continues to evolve, and many exciting things are happening in the project. In this talk we provide an overview of machine learning, PyTorch fundamentals, and the latest developments in the community. We also provide a tour of the many repositories in the PyTorch project, discuss how you can become a contributor, and show how you can leverage the project to create neural network models for your own needs.
