cs702

36 ms·

571.
by cs702 2y ago
The startup raised $2.5M and immediately spent $1.8M of it on the domain name friend.com . That neatly sums up the priorities of many recent "AI" startups.
572.
by cs702 2y ago
It's true that most Olympic athletes are paid nothing, barely make ends meet at home, and are largely ignored by sponsors/advertisers. The OP argues that the reason for this is that the Olympic Committee doesn't do more for a
573.
by cs702 2y ago
> My point is that you're attacking a position that no real critic holds I'm not so sure. I wrote my comment after seeing quite a few comments on other threads here that read like fancy versions of "this is just brute-fo
574.
by cs702 2y ago
Exactly . They are not dumb search!
575.
by cs702 2y ago
That's in fact precisely my point: Clever forms of "dumb search" that leverage compute, DNNs, RL, self-play, and/or formal languages are not dumb at all. I put the words "dumb search" in quotes precisely beca
576.
by cs702 2y ago
In 1997, machines defeated a World Chess Champion for the first time, using brute-force "dumb search." Critics noted that while "dumb search" worked for chess, it might not necessarily be a general strategy applicable to
577.
by cs702 2y ago
When large corporations like banks say they want "access to social media," what they mean is they would love to get their hands on something like this: https://en.wikipedia.org/wiki/Social_Credit_System ...
578.
by cs702 2y ago
Came here to say the same thing, after trying to find a link to the original source, i.e., the purported "study" mentioned in the headline. I was unable to find it.[a] This doesn't deserve to be on HN. --- [a] I mean a proper
579.
by cs702 2y ago
If the model is already pretrained, there's no need to pretrain it, so the cost of pretraining is zero.
580.
by cs702 2y ago
> We’re releasing Llama 3.1 405B, the first frontier-level open source AI model, as well as new and improved Llama 3.1 70B and 8B models. Bravo! While I don't agree with Zuck's views and actions on many fronts, on this occa
581.
by cs702 2y ago
I'm reminded of the classic cartoon, "On the Internet, no one knows you're a dog."[a] On HN, you don't know if you're talking to Satya Nadella. --- [a] https://en.wikipedia.org/wiki/On_the_
582.
by cs702 2y ago
To the moderators: This link should be the OP. The reddit discussion is full of, ahem, low-quality content.
583.
by cs702 2y ago
Wait, what !? If it's true that the company has been intentionally blocking users whenever they try to follow or view media on the Harris account, then... shame on the company . But blocking users from following or viewing media by
584.
by cs702 2y ago
Thank you for taking the time to update everyone on HN. I've added your work to my reading list.
585.
by cs702 2y ago
Surely this story will be turned into a movie that will "bear" no resemblance to the facts.
586.
by cs702 2y ago
Huh, interesting. According to the authors, these "parameteric matrix models" or PMMs outperform: * commonly used (zero- or low-parameter) regression models like XGBoost, random forests, kNN, and support vector machines on a varie
587.
by cs702 2y ago
Yeah, it wasn't cheap. It's still not enough.
588.
by cs702 2y ago
CrowdStrike is a textbook example of a single point of failure: https://en.wikipedia.org/wiki/Single_point_of_failure
589.
by cs702 2y ago
For most corporations, security and robustness are -- and for a long time have been -- an afterthought. Making systems hard to hack and robust to rare events : * is really hard, * costs a lot of money, and * reduces earnings in the short
590.
by cs702 2y ago
A textbook example is corporations that standardize on a single application or service across all infrastructure for a particular functionality. Like, say, using CrowdStrike everywhere to secure all cloud workloads.
591.
How Complex Systems Fail (2000)(researchgate.net)
2 points by cs702 2y ago|0 comments
592.
Single Point of Failure(en.wikipedia.org)
1 points by cs702 2y ago|1 comments
593.
by cs702 2y ago
Good work by well-known reputable authors. The gains in training efficiency and compute cost versus widely used text-encoding models like RoBERTa and XLNet are significant. Thank you for sharing this on HN!
594.
by cs702 2y ago
It's hard to make a fair comparison, because there hasn't been anywhere near as much money, effort, or compute invested in trying to scale up routing methods. Promising approaches are often ignored for reasons that have little to
595.
by cs702 2y ago
The OP is not about evolutionary algorithms in the usual sense (random mutation and selection over many generations). It's about mechanisms in the brain that plausibly evolved over time.
596.
by cs702 2y ago
The short answer as to why capsule networks have "fallen out of fashion" is... Transformers . Transformers came out at roughly the same time[a] and have proven to be great at... pretty much everything. They just work. Since the
597.
by cs702 2y ago
Big-picture, the idea is that different modalities of sensory data (visual, olfactory, etc.) are processed by different minicolumns in the brain, i.e., different subnetworks, each outputting a different firing pattern. These firing patterns
598.
by cs702 2y ago
Apparently, the presence of pirates is correlated with global cooling too: https://doctorspaghetti.org/pastafarians-pirates-and-climate... Harrr!
599.
by cs702 2y ago
Paper: https://arxiv.org/abs/2407.02678 Previous discussion: https://news.ycombinator.com/item?id=40899309
600.
by cs702 2y ago
Wow, this looks amazing . The authors have developed Cradle, a multimodal-LLM-powered agent framework with six modules: Information Gathering, Self-Reflection, Task Inference, Skill Curation, Action Planning, and Memory. Once Cradle has pr

More ›