Sabitlenmiş Tweet
i tried having a balanced life, you know, the whole sleep for 6-8 hrs, have work-life balance etc.
but now, after looking at the kind of changes that are coming up in generative AI, i give up
i give up on being calm 🫡
English
jahnavi yelamanchi
1.9K posts
jahnavi yelamanchi
@sugoi_loop
i solemnly swear that i am up to no good
underground Katılım Ocak 2022
144 Takip Edilen289 Takipçiler
oh the happiness that comes with downloading the manga you cannot find anywhere
English
This guy looks like the surface of Mars. Intelligent design or evolution?
English
new sticker just dropped
Ahmad@TheAhmadOsman
new favorite meme dropped will quote rt papers with it
English
- current architectures can't simulate and compare various chains of thought/reasoning, which is core to intelligence [maybe analogous to the halting problem?]
- world modeling could help by letting models internally simulate and self-correct instead of running expensive rollouts
English
thoughts after the @karpathy interview:
- the limiting point is that models lack the ability to come up with self-criticisms during problem-solving
- people generally have 'what ifs' in their head and dream up alternative paths irl, not after the entire process is done
Dwarkesh Patel@dwarkesh_sp
The @karpathy interview 0:00:00 – AGI is still a decade away 0:30:33 – LLM cognitive deficits 0:40:53 – RL is terrible 0:50:26 – How do humans learn? 1:07:13 – AGI will blend into 2% GDP growth 1:18:24 – ASI 1:33:38 – Evolution of intelligence & culture 1:43:43 - Why self driving took so long 1:57:08 - Future of education Look up Dwarkesh Podcast on YouTube, Apple Podcasts, Spotify, etc. Enjoy!
English
its out gais!!
Dwarkesh Patel@dwarkesh_sp
The @karpathy interview 0:00:00 – AGI is still a decade away 0:30:33 – LLM cognitive deficits 0:40:53 – RL is terrible 0:50:26 – How do humans learn? 1:07:13 – AGI will blend into 2% GDP growth 1:18:24 – ASI 1:33:38 – Evolution of intelligence & culture 1:43:43 - Why self driving took so long 1:57:08 - Future of education Look up Dwarkesh Podcast on YouTube, Apple Podcasts, Spotify, etc. Enjoy!
English
love starting my day with just enough progress to feel totally lost but not enough to fix anything
English
bread crumbs and croutons ≈ red carbon photons 🤦♀️
~ a sleep deprived coworker while fixing the ufactory xarm
English
jahnavi yelamanchi retweetledi
Generalization needs data. But data collection is hard for precise tasks like plugging USBs, swiping cards, inserting plugs, and keying locks. Introducing robust, precise VisuoTactile Local (ViTaL) policies: >90% success rates from just 30 demos and 45 min of real-world RL.🧶⬇️
English
jahnavi yelamanchi retweetledi
Tactile sensing is gaining traction, but slowly. Why? Because integration remains difficult. But what if adding touch sensors to your robot was as easy as hitting “print”?
Introducing eFlesh: a 3D-printable, customizable tactile sensor.
Shape it. Size it. Print it. 🧶👇
English
@knowrohit07 teaching someone how to run a company is a joke anyway; if you don’t grasp it intuitively or learn through trial and error, then just don’t do it in the first place
English
@joodalooped this is nothing compared to the mess in photos app
English
jahnavi yelamanchi retweetledi
jahnavi yelamanchi retweetledi
jahnavi yelamanchi retweetledi
