Sherpa UPDATE - 7 JULY 2023

In This Page:

Sherpa UPDATE - 7 JULY 2023#

Amir Feizpour

Summary#

Amir Feizpour discusses a community project that aims to transcribe and summarize workshop discussions to create an open book accessible to a wider audience. The project utilizes large language models to summarize discussions and make them more accessible. The presentation also introduces the concept of knowledge ops and discusses the creation of an agent called Sherpa. The roadmap for the project includes an alpha release in August and integration into two communities by September.

Topics:#

Community Project: Open Book
  • The project aims to transcribe and summarize workshop discussions to create an open book accessible to a wider audience.

  • Large language models are used to summarize discussions and make them more accessible.

  • The open book is growing rapidly, so the project aims to create a system where people can ask questions and get answers from the open book and other knowledge sources.

  • Contributions to the project, such as adding to the open book, participating in the development of the Slack app, and providing feedback, are encouraged.

Knowledge Ops
  • Knowledge ops involves the management and optimization of knowledge assets and processes.

  • The project aims to automate and augment knowledge processes using generative tools and large language models.

  • Language reasoning skills and tool skills are crucial in enabling knowledge ops.

  • The project uses language models to generate soft questions and subtasks based on user queries, fetch information from different sources, and provide answers.

Agent: Sherpa
  • Sherpa is an assistant or thinking companion that helps users solve complex problems without replacing them.

  • The project involves the development of a tool called OpenBook, which makes event transcripts and workshop notes more accessible.

  • OpenBook automates the summarization process, making event transcripts easier to read and providing a more coherent and concise representation of the content.

  • The project also focuses on creating a knowledge base that indexes various resources, including live books, documentation, and community information.

Roadmap and Collaboration
  • The roadmap for the project includes an alpha release in August and integration into two communities by September.

  • The goal is to have the general availability of the knowledge base, or the ‘hive mind,’ by the end of the year.

  • Contributions of any kind, from documentation to coding, are welcomed.

  • Companies interested in implementing large language models on their internal data can reach out for a discovery call and assistance with implementation.