Humans, with their prior knowledge and their highly general cognitive abilities, handle this just fine :)
Show this thread
To be clear, if each ARC task came with 100,000 examples, you could most likely train a DL model on each task and shove ARC that way. But they don't -- you get 2-4 examples per task. So now you have to exhibit some intelligence, not just fit a curvehttps://twitter.com/fchollet/status/1234789789309652992 …
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Yes, exactly. One of your train tasks contains just 2 examples and there are at least 4 simple and obvious rule and patterns that predict the output. Humans are able to narrow that down from patterns learned from other tasks, current ARC solvers, not so much
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.