Thoughts on what Sutton called the Bitter Lesson? http://incompleteideas.net/IncIdeas/BitterLesson.html …
I worked on scaling up existing ML algorithms until I found that even with infinite data they'd still be too stupid. Since then I've focused on learning smarter before learning bigger, and I bet that sooner or later the whole field will have to do the same.
-
-
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
The key is imitation.
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
You know what that might not be entirely true, it definitely works in some cases, speech recognition for instance
-
It works in most cases, but then it plateaus, and the question is what happens after.
- Näytä vastaukset
Uusi keskustelu -
-
-
Current DL/ML methods are a hack at best. We pushed them to their limits and they sorta work for but it's like trying to go to the moon with a bicycle. Numenta is onto something, more people should pay attention to their work objectively.
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
What algorithms are you considering smarter?
- Keskustelun loppu
Uusi keskustelu -
-
-
So you disagree with Jack Nicklaus, who said that first you had to learn to hit the ball a long way, *then* you had to learn to hit it straight...
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
Do you think there are some models that are still not scaled up enough to be considered too stupid? I.e. is GPT-n like that? Can we for sure say it digested enough data to give up on it ever learning smart? Is there a research on boundaries?
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
Thinking out out - Any thoughts on this? https://vigorotaku2021.blogspot.com/2021/09/autonomous-vehicles-2021-how-do-we-get.html …
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
Lataaminen näyttää kestävän hetken.
Twitter saattaa olla ruuhkautunut tai ongelma on muuten hetkellinen. Yritä uudelleen tai käy Twitterin tilasivulla saadaksesi lisätietoja.