@kimmonismus And deepseek training cost was 5 million

RT @UPSC_Notes: Nothing bro just filling my UPSC application form in 2025

@LostHero5 @MuradMAlqadi @burkov @Paulfruitful_ Yet

@burkov ML models don’t just regurgitate training data. AlphaZero learned purely through self-play, discovering novel strategies like the Shogi Gold Move & dynamic piece values—proving ML can generate new knowledge beyond its training data."

@Gilbertocintro4 Fleas have smaller fleas and so on to infinity! to infinity ?

@dwarkesh_sp LLMs operate in token-limited bursts and lack long-term memory or self-feedback loops, unlike humans' unlimited continuous thought input. This limits deep connections and hinders true novel discovery.

@mmay3r LLMs operate in token-limited bursts and lack long-term memory or self-feedback loops, unlike humans' unlimited continuous thought input. This limits deep connections and hinders true novel discovery.

@kimmonismus so do deepseek

@IsaacKing314 Exactly that's not how math works

@ESYudkowsky Can LLMs lead to AGI? If so, when (assuming current progress)?

The geniuses who cracked X algorithms, don’t keep all the secrets—drop some wisdom and let us level up too

@BNODesk we need ASI ASAP

@DaveShapi that's what i have been thinking that's the true life we meant to live and ASI will enable us to live that life

@kanyewest

Did eric cartman hack kanye twitter or what

@anammostarac Including Issac Newton

@ThePhysicsMemes You still using 4o

🤯

@ThePhysicsMemes 🤯

@chatgpt21 nah he's a hype man that's his job