nothing beats a little retreat in the countryside, and a fresh harvest of pears and black currants.
217
LeCun's argument against llms
Replying to @IntuitMachine
i was at the conference and it’s main thesis is
- as they are predicting tokens after tokens they can’t be reliable for very long
- they can’t make a world model
and some other arguments about bits of information humans go through vs the amount of bits of data llm have been fed
495
David Nizovsky retweeted
Groundbreaking technology often starts as "a solution looking for a problem." The laser was once dismissed this way, yet transformed countless industries.
You don’t need immediate applications, just look for transformative potential.
my favorite podcast about quantum so far
Episode with Misha Lukin talking about cold atoms, lasers and quantum computers is in the air.
youtube: youtu.be/ewsPxTGKw30?si=pbXN…
spotify: open.spotify.com/episode/769…
2
148
parametrizing qubits (thread)
picture this: you're standing in front of a quantum computer. what a potential. but how do you actually use this tech to solve real world problems? at some point you will need to use something called parametrized quantum circuits (pqs).
1
1
110
the challenges
of course, it's not all quantum rainbows. pqcs face some serious problems:
- barren plateaus
- noise
- scalability
so right now it's like to find a specific grain of sand in the sahara desert. until we have a good quantum computer or a simulator.
1
2
114



















