Jingyao Zhao

[Jingyao Zhao @Zhaoj6317]

the "flu cycle"🥲 stage 2 now

[Jingyao Zhao @Zhaoj6317]

@micahgallen our brain is overparameterized as well? 😯

[Jingyao Zhao @Zhaoj6317]

@VraserX From my observation, actual good stuff would be appreciated. What offends people are creations by someone who lacks the knowledge/understanding, doesn't respect the field, makes a toy from real experts' view, yet feels good about themselves.

[Jingyao Zhao @Zhaoj6317]

@asunokibou solved, not rigorous and not elegant imo. would expect better idea.

[Jingyao Zhao @Zhaoj6317]

@NateLorenzen me try not to be stupid everyday 🤪

[Jingyao Zhao @Zhaoj6317]

@dr_appie that's so cool

[Jingyao Zhao @Zhaoj6317]

@filippkowalski There were indeed great, talented designers, and Adobe indeed didn't behave nicely providing their properties. Acknowledging the first doesn't affect people celebrating regarding the second I guess.

[Jingyao Zhao @Zhaoj6317]

@Moleh1ll Humans simply won't admit they lose.

[Jingyao Zhao @Zhaoj6317]

@arpitrage The whole academic community rewarded being first. Now AI might become the first one that otherwise might be you. More importantly, you at the frontier would need to check and validate for humans without deserved acknowledgment.

[Jingyao Zhao @Zhaoj6317]

@moiiikaaa that would be horrible just listening it 😱

[Jingyao Zhao @Zhaoj6317]

@buildwithsid m3 16gb is here 🤚🏻 mac would behave like a 20 tests old x86 laptop, even codex ui response with constant and obvious latency

[Jingyao Zhao @Zhaoj6317]

@SleevelessRube @zeta_globin I guess so. That would explain why they're still the choice. I'd expect specially modified models for future research.

[Jingyao Zhao @Zhaoj6317]

@cremieuxrecueil indeed he fails to calibrate understanding in real bio/med, but also true that no one would resist knowing the mechanism. while blaming the wrong metaphor, we should acknowledge the spirit behind it IMO

[Jingyao Zhao @Zhaoj6317]

@mathepi @imjustnewatai Yeah, mathematicians would learn like how Go players learned from AlphaGo years ago. And finally we improve.

[Jingyao Zhao @Zhaoj6317]

@shawmakesmagic it's about the nature of intelligence that no one, as far as I know, could tell for now. same logic could be applied to brains as well: human brain is conscious; smaller brains, say dogs, less but still kinda; even smaller, say insects, remain meaningful behaviors, ...

[Jingyao Zhao @Zhaoj6317]

@Helios_Hua would this rely on manual definition and specially conditioned & collected data? it would feel more supervised and harder to scale if so. fantastic and inspiring work anyway 🥳

[Jingyao Zhao @Zhaoj6317]

@Salmo_Lab cool visualizations are always the best part

[Jingyao Zhao @Zhaoj6317]

@yoheinakajima @vertinski could anyone explain the problem and why the shown solution is good?

[Jingyao Zhao @Zhaoj6317]

Please share one insight about JEPA-style models that impresses you the most🎙️: