Aaryan Kakad·6d agoWhy you don’t need more intelligent models anymore?So, I asked GPT 6.1 to research a bunch of 2025 frontier models and their benchmarks and compare them with 2026 open-weight models.A response icon1A response icon1
Aaryan Kakad·Nov 5exploring the back propagation paperthis is a raw and random exploration write-up, where I go through the back propagation paper and write what I feel in my own words.
Aaryan Kakad·Oct 24how to self learn anythingingest, build, hit a wall, fill the gaps, repeat. forever.A response icon1A response icon1
Aaryan Kakad·Oct 20tesla used vacuum cleaners and rockets in a car to break physicstesla made their new roadster go from 0 to 60 mph in 0.81s. here’s how they made it happen.A response icon1A response icon1
Aaryan Kakad·Oct 16can two AI agents communicate without using a language?when I want to say something to you, I have no choice but to use a language. but why are we forcing agents to do the same?A response icon1A response icon1
Aaryan Kakad·Oct 6recursive self improvement (explained from scratch)just imagine the language model you use becomes capable of improving itself recursively, wouldn’t it be dangerous?A response icon1A response icon1
Aaryan Kakad·Oct 3how a computer works (explained from scratch)this article is all about decoding how a computer works behind the scenes.
Aaryan Kakad·Sep 25google is about to setup a DATACENTER in the SPACE (here’s the physics behind it)google is planning to send tpu chips to space for unlimited power - project suncatcher.A response icon2A response icon2
Aaryan Kakad·Sep 22pace the frontier? no.you can read the article for free here if you aren’t a medium member…
Aaryan Kakad·Sep 10deepseek v4.1 flash’s alien architecture explained.890 bytes is all deepseek v4.1 flash needs to store one token.