We are pleased to share that using Gauss, we have completed a ~200K LOC formalization of Maryna Viazovska’s 2022 Fields Medal theorems on optimal sphere packing in dimensions 8 and 24. This is the only Fields Medal-winning result from this century to be completely formalized,

We initially thought GLP-1s like Ozempic, Tirzapeptide and Retatutride just reduced food cravings. Now, we know they work for alcohol, cocaine, gambling and other addictions too But do you know what runs on exactly the same circuit? Falling in love GLP-1 receptors sit in the

:d

saas will die ... let's see how github.com/kronael/kanipi manages to replace a host of it for small individual agents.... and individual agents will be small don't be fooled

so...the nanoclaw is a bit of a scam ... its small all right but there's really nothing there

we'll relive the which language is the right language again... now just with an ai tinge ... buckle up functional bros

most likely just click bait but i choose to click anyway

New paper out: AI Must Embrace Specialization via Superhuman Adaptable Intelligence With Judah Goldfeder, Philippe Wyder, and Yann LeCun . There is quite a lot of buzz on our paper, so here is my take. Everyone's talking about AGI, but nobody agrees on what it means, and that

solid work ... focus will be the most valued resource ... some will just vibecode into oblivion ... some will start to do statistical data learning

I don’t think you can TACO tweet out of refineries being blown up tbh

The current highest level 3 score was achieved by Nvidia, leveraging a multi-agent system that includes Nvidias own tool orchestrator model. It scores 89.8% on Lvl 3 (even higher than the 88.9% typo I wrote above) The public leaderboard can be seen here: huggingface.co/spaces/gaia-be…

someone will apply this to trading. this will be your opportunity anon to frontrun what they discover and be one step ahead...

good that github.com/kronael/kanipi is rather a deployment and organization leveraging an underlying tool call system. currently Claude Code. you can use it to design pretty much any agent... check it out at krons.fiu.wtf/kanipi

Handling a swarm of agents will be pretty soon doable using github.com/kronael/kanipi ... on discord or slack. But mostly importantly other workflows unimaginable today thanks to its refreshing approach to minimalism, convention over configuration, and safe agent-extendability

with github.com/kronael/kanipi youre able to ship whole products not just personal assistants in an extensible and tractable way... also using the agents but because prevented from the slop ... its maturing but will be very much complete soon

Here it is! Session Stripper - a skill that does surgical JSONL session stripping of your Claude conversations Remember when I reverse-engineered Claude Code's binary to add /microcompact? A lot of you asked for the same thing but without patching the binary Now you can do it!

During the 30-minute container session on March 7 at ~21:12, the mayai agent (running in the WhatsApp "May AI" group as root) called the schedule_task MCP tool to create a daily cron job. It wasn't explicitly requested by anyone in chat — the conversation was about web

I too have a sentient agent now. Finally, I almost thought I will never get to have one...

intelligence might be a measure on how fast you adapt with scarce data ... then LLMs are dumb as _

we are now all just exploring the internet where the URL is the infinite prompt ... quite thrilling what we'll find ... recombined