cma

51 ms·

931.
by cma 2y ago
I don't like it, but how would you prevent everyone from getting expensive schooling and then immediately declaring bankruptcy? Just better redistribution and georgism/UBI type stuff but also keeping the need based stuff (medicaid
932.
by cma 2y ago
I can't imagine getting assigned this task and not getting depressed, "bullshit jobs" smacking you in the face
933.
by cma 2y ago
I doubt they prioritized it. To get normal maps you usually first need a high resolution mesh, but then need other steps to get good decimation for lods and normal bake. That's mostly extra work, not alternative work that wasn't
934.
by cma 2y ago
First it was AI articles, raising it to entire successful book trilogies seems like a much bigger leap. Even considering the largest context windows they wouldn't directly fit and there is much less data to train context of that size o
935.
by cma 2y ago
Can constructors and destructors still go counter to this idea if needed?
936.
by cma 2y ago
Ironic how this comment exemplifies the issue - broad claims about "slop" output but no specific examples or engagement with current architectures. Real discussions here usually reference benchmarks or implementation details. (fro
937.
by cma 2y ago
Also challenging aspects of their explanations to get at something better is good for developing critical thinking.
938.
by cma 2y ago
> The other thing was that o1 had access to many more answer / search strategies. For example, if you asked o1 to summarize a long email, it would just summarize the email. The full o1 reasoning traces aren't available, you jus
939.
by cma 2y ago
Earliest I've seen with 'Eh?' as an interpreter response is RAND's JOSS: https://en.wikipedia.org/wiki/JOSS#/media/File:JOSS_Session.... https://en.wikipedia.org/wiki/
940.
by cma 2y ago
> I just saw Sam Altman speak at YCNYC and I was impressed. I have never actually met him or heard him speak before Monday, but one of his stories really stuck out and went something like this: > "We were trying to get a big clie
941.
by cma 2y ago
OpenAI's benchmark results looking like Musk's Path of Exile character..
942.
OpenAI has the FrontierMath dataset; the mathematicians involved were unaware(old.reddit.com)
2 points by cma 2y ago|0 comments
943.
by cma 2y ago
Facebook is not being banned from China. China just asks that Facebook sell itself to a non-American company or they will be banned (but it's not a ban). Any of those opinions China is "suppressing" by not allowing Facebook
944.
by cma 2y ago
It's their legal out against having to research stuff to prevent libel liability. And they can embed social media photos and videos that weren't even from the rights holder to avoid having to clear rights to anything.
945.
by cma 2y ago
That can still happen without Chinese ownership too.
946.
by cma 2y ago
Sounds very similar to the great firewall
947.
by cma 2y ago
They weaponized the backend to drive that by injecting in recommendations for that inorganically, or that was something that emerged from having a recommender system backend?
948.
by cma 2y ago
Falcons upper stage isn't recovered or refurbished at all. And sometimes it launches the whole thing in expendible configuration more like SLS's use of shuttle engines than shuttle's. 30-40 years after shuttle yes we have imp
949.
by cma 2y ago
It would be more like Coke was Mexican owned and HFCS was outlawed in Mexico. Then Mexican Coke used sugar and the Coke they exported to America used HFCS. And America said, hey, you're not consuming the same Coke you send here: we&#x2
950.
by cma 2y ago
I think it's widely agreed a lot of useful stuff came out of Schmidhubers lab. The example I gave was one of the first things that scaled in lots of ways especially in depth, and it shares some characteristics with this. I doubt it out
951.
by cma 2y ago
The whole country had a shift though, they implemented gaming and entertainment regulations and video sites like bilibili went from $153 to a low of $8 a share.
952.
by cma 2y ago
"I like the idea of a meta-mechanism that learns to update an associative memory based on how surprising the data is." Just pointing out that that idea was in some of Schmidhuber's earlier work. > Schmidhuber (or in this c
953.
by cma 2y ago
The SRBs could land in the ocean with parachutes and be recovered and refurbished. Shuttle wasn't economical as I mentioned, and definitely the space shuttle wasn't safe. What you claimed was: "They have the first rocket eve
954.
by cma 2y ago
> They have the first rocket ever made which can take payloads to orbit and then be reused. The space shuttle did this over 40 years ago. You can argue SpaceX have the first economical one 40 years later, but the second stage isn't
955.
by cma 2y ago
1991 > Each RNN tries to solve the pretext task of predicting its next input, sending only unexpected inputs to the next RNN above. This greatly facilitates downstream supervised deep learning such as sequence classification. By 1993, th
956.
by cma 2y ago
500K context on enterprise Claude
957.
by cma 2y ago
The Lenovo Legion Go I think has this for FPS games. Joycon like thing goes into a little slider adapter and becomes a vertical mouse.
958.
by cma 2y ago
Sure, just saying it wasn't an impossible feat.
959.
by cma 2y ago
The space shuttle achieved reusable booster landing in the 80s with parachutes and water. Delta clipper controlled burn relanding in the 90s but not scaled to orbital class.
960.
by cma 2y ago
> with a few exceptions 20% of the population is exempt, unless you are saying Arab Israelis aren't Israeli. More including many orthodox were also exempt up until mid 2024.

More ›