- Published on
Gemini 1 million
- Authors
- Name
- Martin Andrews
- @mdda123
Presentation Link
Sam Witteveen and I started the Machine Learning Singapore Group (or MLSG for short) on MeetUp in February 2017 (it was previously named "TensorFlow and Deep Learning Singapore"), and held the ~fiftieth MeetUp! on 20-Feb-2024. Actually, we'll probably celebrate 'officially' at the next one - since that will mark the fiftieth MLSG event at which we've both spoken...
The overall event was re-titled the "Gemini 1 Million" MeetUp because Google announced the 1 million token context window Gemini 1.5 model during the previous week - and both Sam and I demonstrated it during and after the event.
My talk "Self-Improving LLMs", and looked at the following topics:
- DeepMind's Alpha-* papers
- And a recent improvement : AlphaCodium (non-Google)
- Self-Discover paper
- ... including a demo using Gemini 1.0 on VertexAI (Google Cloud)
- Direct link to Colab implementation
- Plus a look at a key element : PromptBreeder
- ... including a demo using Gemini 1.0 on VertexAI (Google Cloud)
Sam talked about his experiments with Gemini 1.5 (mostly through slides, since the generation isn't super-fast).
We also were pleased to have Shubham Gupta giving an in-depth talk titled "State Space Models 101" which included:
- Key elements of State-Space-Models
- Considerations for performance on GPUs
- Implementation of the new Mamba model
In addition, Nicholas Chen gave an interesting lightning talk titled "LLMs + Cognitive Architecture = Generalist Agents?", which covered:
- the CoALA framework for thinking about intelligent systems
- and how recent works like VOYAGER and Generative Agents can be understood
- a new intelligent agent for automating computer tasks "FRIDAY"
Many thanks to the Google team, who not only allowed us to use Google's Developer Space, and put out the extra chairs required for our larger-than-expected attendence, but were also kind enough to provide Pizza for the attendees!
The slides for my talk, which contain links to all of the reference materials and sources, are here :
If there are any questions about the presentation please ask below, or contact me using the details given on the slides themselves.
