BEL-2, SEN-2 OMFG ... what a world cup so far.

View live event →

Going off to the live streaming mines. See you there.  View live event → If you like to open up in your own player: https://laantungir.net/stream/industrial/src/index.m3u8

Trying to figure out why I love this chair so much.

It can get depressing looking at all the censorship/kyc coming our way. So let’s talk about some really positive news. The cost of AI is dropping incredibly fast. AI token costs have been falling 5x - 10x per year. Frontier model cost has been dropping 2x - 4x per year. Much of this cost comes from algorithmic improvements, but even if you look at hardware or inference cost, that has been dropping 1x - 4x per year. So lets say conservatively 50% drop in cost per year. Imagine you’re in a society that wants to ban the printing press, but printing presses cost 50% less every year. It’s hard to ban cheap things. Technology and time is on our side.

GM

GM

@waxwing inspired me from his post about using LLM's for encryption. Here is an example web page that will download GPT-2, and then encrypt a secret message, and decrypt it. The encrypted message looks like regular text, but hidden in the text is a secret message. Clearnet:

LLM STEGANOGRAPHY

Tor: http://kn2jam4kyz6s5wacyozo3do3d2zsvh4to45uh7xliulwdust6zjuvnad.onion/client/llm-steganography.html FIPS: http://npub1crpldvy49ef8z34wlacwujnfudy4nd7k96aqdx5wgn6ckztz7z8q9t59ud.fips/client/llm-steganography.html View quoted note →

Unfortunately, GLM 5.2 still looks pretty hard to run at home. View quoted note →

GLM 5.2 has become my daily driver. View quoted note →

GM. Time to get up.

Since everyone loves monorails.