EXPERIMENT: LATENT LOGBOOK
"The Latent Space is dark. We navigate it together."
Latent Logbook is a collective discovery experiment designed to explore the capabilities of Large Language Models (LLMs).
I’ve always loved the energy of December coding events, but I wanted to try something a little different this year focused on LLMs.
The idea is simple: we all get the same puzzle, but we can solve it using whatever model or technique we want. Then we compare notes. I'm not sure if it will work perfectly, but I think the coolest part will be seeing the five different ways five different people solve the same problem.
OPERATIONAL OVERVIEW
1. The Chapters
The mission is divided into thematic Chapters released weekly.
- Chapter I: The Text (Basic Prompting)
- Chapter II: The Signal (Audio/Transcripts)
- Chapter III: The Vision (Multimodal)
Puzzles unlock in chapters, letting you solve them at your own pace without the daily deadline pressure.
2. The Currency: Latents (⏣)
We use a simple social currency to help the best ideas surface.
- Earn: Receive 5 Latents (⏣) every time you submit a solution.
- Spend: Use Latents to "Power Up" other explorers in the Archive.
Checking every possible valid output from every possible model is technically impossible. We trust that if you're hitting submit, you've actually put in the effort to find a solution.
3. The Archive
Solving a puzzle grants you access to The Archive for that chapter. This is a shared log where you can view the prompts, chat logs, and code scripts used by other explorers. It is a place to compare different approaches to the same problem.
Weather forecast: Heavy rain expected.