Published on

IPhO Gold using Agentic Gemini

Presentation
Authors

The sixty-forth MeetUp of the Machine Learning Singapore Group, was titled : "LLMs in production : Inference, RAG, and video models" - and was a bit of a change for us because it was held at the Rakuten offices near Raffles Place in Singapore (rather than Google's offices).

My Presentation

My talk was titled "IPhO Gold using Agentic Gemini", and had the following outline:

  • Intro + What is the IPhO?
    • Including a quick walk-through of this year's 3 IPhO Theory questions
  • Agentic Gemini for the IPhO
  • Wrap-up & QR-code (the latter to reduce audience distractions)

This was an exciting paper for me to present because I actually represented the UK in the International Physics Olympiad in two different years (before I went up to university). So, it was a bit Physics-oriented, though I think the audience was more interested in the agentic use of Gemini Pro...

In-Person version

The slides for my talk, which contain links to all of the reference materials and sources, are here :

Presentation Screenshot

If there are any questions about the presentation please ask below, or contact me using the details given on the slides themselves.

Presentation Content Example

YouTube version

16 min "AI Wins IPhO Gold" video on YouTube

Other Presentations

The MeetUp included two other speakers (both from Rakuten) - Dongzhe Wang and Jonathan Zhao:

  • Dongzhe gave the first talk "Efficient LLM Fine-tuning for Semantic Search", going from the basic ideas of semantic search to the process of fine-tuning an embedding model for a specific use-case.
  • In Jonathan's talk, "Efficient Inference and Serving of LLMs and Large Video-Generative Models", he explained the key aspects of modern LLM inference serving, and also highlighted some of the other issues in serving video models.

Acknowledgements

Many thanks to the Rakuten team, who allowed us to use Rakuten's centrally located meeting space.