Large language model (LLM) applications like ChatGPT have taken the world by storm. This course will teach you how to leverage the latest developments in LLMs for real-world natural language processing use cases. Through lectures, tutorials, and hands-on labs, you will learn how to develop and deploy your own LLMs, leveraging foundation models such as GPT and BERT. We will cover safety, societal impact, and ethical considerations of LLMs, as well as how to evaluate trained models. This course concludes with best practices for deploying LLMs in production.

By the end of the course, you will:

Understand what LLMs like ChatGPT are from the ground up and use them for real-world business cases.
Use transfer learning techniques, such as few-shot learning, to customize Databricks' open-sourced Dolly for specific tasks.
Apply popular NLP topics such as Transformers, BERT, and GPT effectively on Databricks.
Understand the nuances of pre-training, fine-tuning, and prompt engineering for tuning LLMs.
Create, implement, and evaluate LLMs using popular publicly available datasets.
Articulate best practices for deploying LLMs in production.

In addition to this class, we've also launched a LLM program with edX, which consists of two courses. As there may be significant content overlap, we recommend that you dive into the additional learning content on edX following the conference to earn an LLM Certificate.

Prerequisites:

You will need to bring your own computer and connect to the USC wi-fi network to be able to participate in this workshop;
Intermediate-level experience with Python;
Working knowledge of machine learning and deep learning is helpful.

Please note that this session will not be recorded and will not be available online at a later date.

SGM 124

Sessions in SGM 124