malloc

opus 4.7 is so lazy

people don’t understand this take cause they don’t understand what’s happening in AI memory. Everything is moving to git backed files accessible via grep-type-systems or semantic plus grep which isn’t very defensible to offer as a service. In other words… the SOTA approaches to memory are now just agent plus terminal. And all the fancy approaches like knowledge graphs are getting rekt by an agent plus a terminal. Your fancy agent structure is getting rekt by a model that can keep track of anything over 1000+ terminal calls.

thought about writing and this and making life align with what you want to write about & vice versa...

@mypacetsuyoshi <3

Thank you too. I'm glad you liked it.🙏

Hyper Connection + Looped Transformer.

man sometimes i forget how early i was sometimes to the game but not too early just like ~2021, before that i planned on making video games (still do)

thanks you @mypacetsuyoshi for this photo unsplash.com/photos/a-pond-…

A tragic SF archetype you’ll see a lot is someone who has a virtuoso’s grasp of some passion in the humanities, but they muzzle it & flagellate themselves into ‘actually really liking’ agentic scaffolding or whatever

there goes a whole yc batch

the stupid bat paper I stg

Anthropic is obliterating OpenAI Claude Mythos 77.8% on SWE-Bench Pro 20% higher than GPT-5.4-xhigh

timely

zürich today and i love swans & bob dylan and his album bringing it back home which is when he first went electric in Greenwich village at the age of 23

@moonsandhues I never really liked carbs as much as the actual protein in most dishes (usually meat ig although I’ve been less carnivorous since moving to Europe) but I force myself to eat them

carbmaxxing vs proteinmaxxing ????

I decided to write down some of my thoughts in blog posts. In this post: zhaomingxie.github.io/blog/posts/wor…, I share my thoughts on world model, purely in the context of using simulation data to train a dynamic model of an articulated rigid body system, e.g., a humanoid robot.

@MeirYossef_Levi proceedings.mlr.press/v119/wang20k/w… Reminds me of this, nice investigation

Why does contrastive learning produce Gaussian representations? My colleague - Roy Betser shows it’s not accidental, just geometry + probability. A simple perspective on InfoNCE from our #ICLR2026 oral presentation. Worth a read: 👉 @rbetser93/why-contrastive-learning-naturally-produces-gaussian-representations-bc4058a8d46c" target="_blank" rel="nofollow noopener">medium.com/@rbetser93/why…