Adapting Large Language AI Models for GEOINT Applications 

Event Time

Originally Aired - Monday, May 6 7:30 AM - 8:30 AM

Info Alert

Create or Log in to My Show Planner to see Videos and Resources.

Info Alert

Your account does not have access to this session.

Videos

Resources


{{video.title}}

Log in to your planner to join the zoom meeting!

{{chatHeaderContent}}

{{chatBodyContent}}

Session Recording

Info Alert

This Session Has Not Started Yet

Be sure to come back after the session starts to have access to session resources.

Event Location

Location: Sun Ballroom B


Event Information

Title: Adapting Large Language AI Models for GEOINT Applications 

Description:

Training summary: ChatGPT and similar large language models have become popular for a variety of applications based on the broad swath of knowledge harvested from the internet. Adapting such models to the specific needs of a GEOINT analyst remains a challenge. As an example, we will use the domain of simulated medical triage as a specialized application where the pretrained large language models have some but not a lot of background knowledge and expertise. We will leverage our work from the DARPA In the Moment (ITM) program to provide motivating examples, tools, and results from this domain. We will demonstrate how to adapt and refine some of the recent large language models with the text documents from the domain, for example the TCCC medical triage guides. Furthermore, we will also demonstrate an approach to adapt the AI model’s decisions to be aligned with those of human decision-makers. Finally, we will also provide other potential adaptation examples that may be relevant to the GEOINT users.

Learning outcomes: Attendees will: 

  • Be able to better understand the strengths and weaknesses of different large language models (LLMs) like ChatGPT, Bard, etc. These models have played a vital role in the state of the art in the field of generative AI.  

  • Be able to also understand other related generative models for other modalities including imagery and speech. For instance, the Stable Diffusion generative models have been popular for generating images of various types including downward-looking satellite images that are very relevant to many GEOINT users.  

  • Be able to familiarize themselves with the tools and techniques involved in adapting the LLMs for more specialized tasks that may be relevant to different GEOINT users.

Prerequisites: While not required, it would be beneficial to have used some of the models like ChatGPT, Bard, and others to have established some familiarity with the nature of these chat systems. This will provide some exposure to the strengths and weaknesses of such AI systems and will make this training more interactive and a productive experience for all parties involved. We will utilize tools like Llama Index (https://www.llamaindex.ai/) in this training among others. Therefore, it will be helpful if the learners have already looked into such tools to get some familiarity with providing additional knowledge in your constrained environment to these LLMs. 

Type: Training


Notes

Create or Log in to My Show Planner to add notes.


Speakers


Tracks