Re:Astra can read and write weights to improve Qwen
by XCSme on Monday November 02, 2026 @04:55PM
(#50344405)
Good luck with figuring out what it does and enforcing any safety switches and rules I guess...
Re:Astra can read and write weights to improve Qwen
by __MatrixMan__ on Monday November 02, 2026 @05:02PM
(#50344495)
It doesn't seem like there's room for recursion here though. A smart model can share its smarts with a dumb one, but the ceiling remains unchanged. Perhaps there's juice to squeeze in verifiable dimensions re: software, math, but there are similar ceilings to be found in those places. We can't keep asking harder and harder questions infinitely while remaining able to verify the answers.
It is astounding, but I don't think it's well set up for anything besides faster approaches to existing limits.
Re:Astra can read and write weights to improve Qwen
by FuckButtons on Monday November 02, 2026 @10:07PM
(#50347536)
> We can't keep asking harder and harder questions infinitely while remaining able to verify the answers.
The last several millennia of mathematical research would seem to be an obvious counter example to this claim.
Re:Astra can read and write weights to improve Qwen
by __MatrixMan__ on Tuesday November 03, 2026 @12:18AM
(#50348381)
Perhaps we have different things in mind re: hardness. But to set that aside, what I should've said is that we can't do this quickly enough for it to be seen as an inexhaustible supply of training material. The consumer and producer are just not well matched.
Re:Astra can read and write weights to improve Qwen
by solid_fuel on Monday November 02, 2026 @05:11PM
(#50344625)
> This is the "recursive self-improvement" they've been warning about.
Well, except for the self part since it improved the performance of a smaller, dumber network.
And the recursive part, since it couldn't improve the performance of a larger network.
[dead]
Re:Astra can read and write weights to improve Qwen
by solid_fuel on Tuesday November 03, 2026 @01:33AM
(#50348772)
I’m certain the next big frontier model will also be able to improve the weights of comparatively tiny models. That does not improve the large frontier model.
Re:Astra can read and write weights to improve Qwen
by cman1444 on Tuesday November 03, 2026 @01:39AM
(#50348802)
Only the context window, right? Which seems....not that big of a hurdle to overcome or navigate around.
Re:Astra can read and write weights to improve Qwen
by danielmarkbruce on Monday November 02, 2026 @09:21PM
(#50347174)
No, no it isn't.
I've always laughed at that phrase, thinking "haven't these things been improving themselves since the very beginning? This is all just iterative gradient descent, each step is a 'self-improvement'", but this really is different.
I wonder if we'll ever discover a new branch of linear algebra or something that allows us to find the optimal weights much more efficiently? Maybe AI will help us discover it?