lil project update: graduated to make a cute personal sidekick with useful tools so you can reclaim your life! Go on walks, capture ideas, take photos, listen to music, set timers, have a morning alarm, *no phone required* A personal computer that helps you be more present!

Solving NYT Connections with 100% accuracy is the ultimate test of AGI (I am AGI)

I always learn a lot more from in-depth analysis of few random cases over dashboards of aggregate statistics across all cases. Both projections can be helpful but the latter is disproportionately pervasive.

Introducing Babbl, a Python library that turns your Markdown files into beautiful .html - Easily render functions and classes live from the codebase - Embed .html (e.g. interactive plotly graphs) - The .md files live inside your git repository and are version controlled -

We were able to reproduce the strong findings of the HRM paper on ARC-AGI-1. Further, we ran a series of ablation experiments to get to the bottom of what's behind it. Key findings: 1. The HRM model architecture itself (the centerpiece of the paper) is not an important factor.

All the env + RL startups depend on ASI not happening and AI investment continuing to grow exponentially

I basically think the one-step grad trick used by HRMs is just RL-ified diffusion

Little update about what I've been cooking up Say hi to Kami A cute, colorful AI sidekick that lets you - leave your phone behind - organise and talk to your context - proactively nudges you towards who you want to be And it can live on your devices or our hardware A WIP!