LLM Course: Week 6 on Use cases and applications of LLMs
Last week, I had a pleasure to teach the Week-6 topic: “Use cases and applications of LLMs”. Week-5 on RAG can be found here.
We looked at multimodal LLMs, as a very interesting, and in many ways still an emerging trend in the LLM world, covering text, image, video and audio modalities (you can ask: “What do you hear in this video?”, for example, — besides a more regular “Does the man wear glasses?”). We covered PaliGemma, ColPali, Video-LLaMA models.
I’ve also demoed various industry-grade LLM applications, including:
* TomTom’s Intelligent Vehicle Assistant — shoutout to Massimiliano Ungheretti, PhD (who starred in the lecture) and IVA team Ola Aleksandra Ferdzyn-Grzelak, Alexey Samuleenkov Cezary Draus and the rest of the team
* BabyAGI — built by a VC, showing the power of modern LLMs to solving multihop challenges, like assembling a list of under-represented topics for a blog post
* Code assist (GitHub Copilot, Cursor) and “office-assist”
* Multimodal vector search we have built with Aarne Talman, PhD for the GSI Technology
In the end we had a fun group brainstorming session, where students came up with the fields deserving to be improved by LLMs, benefits, risks and their mitigation. Sessions like this sparkle creativity in the audience and turn the class around (students help me complete the slides, which is always cool and appreciated).
In the lab we looked at:
1) querying tables (extracted from PDF files) using natural language: “What was Google’s operating margin for 2024?” -> “According to the table, Google’s operating margin for 2024 is 32%.”
2) generating synthetic data
Link to the YT recording:
Follow the podcast for more content and interviews around LLMs, search, RAG, and emerging trends.
