nickpsecurity

9 ms·

91.
by nickpsecurity 8mo ago
Nvidia released Parakeet which claimed superiority. Doesn't negate your point but I did want to add it.
92.
by nickpsecurity 8mo ago
It can be 10,000 requests a day on static HTML and non-existent, PHP pages. That's on my site. I'd rather them have Christ-centered and helpful content in their pretraining. So, I still let them scrape it for the public good. It h
93.
by nickpsecurity 8mo ago
There were a number of methods for doing TLA-like stuff. Others included SPIN/Promela, Pi Calculus (IIRC), and Event-B.
94.
by nickpsecurity 8mo ago
The author has some good points. I'll highlight a few: 1. C language is small enough for one book. 2. It's been described, implemented, and fixed so many times on the Internet that the pretraining data is full of this. 3. I'l
95.
by nickpsecurity 8mo ago
You and I commit sins, too. Should everything we've ever said be disregarded? If you are sharing opinions, you can't possibly believe that. If you expect people to listen to and weigh your comments, we should consider his as well.
96.
by nickpsecurity 8mo ago
God's Word says all have sinned and fall short of the glory of God. That we have choices but all choose evil. We keep choosing evil at times out entire lives. So, all people are to face justice for their evil. But, God so loved the wor
97.
by nickpsecurity 8mo ago
The Bible tells us how to pick godly leaders. It also gives many examples of those doing right and wrong. For both, leaders exist today with similar worldviews. So, it's directly applicable. Thomas Jefferson took inspiration for our sy
98.
by nickpsecurity 8mo ago
I think one could also use a subset compatible with a formal semantics of C. Maybe the C semantics in K Framework, CompCert C, or C0 from Verisoft. Alternatively, whatever is supported in open-source, verification tooling. Then, we have bot
99.
by nickpsecurity 8mo ago
It could be difficult. My first thought would be a SELinux policy like this article attempted: https://danwalsh.livejournal.com/28545.html One might have different profiles with different permissions. A network service usua
100.
by nickpsecurity 9mo ago
Ok, let me give you examples. Much of what many learned about life came from their parents. That included lots of foundational knowledge that was either true or worked well enough. You learned a ton in school from textbooks that you didn&#x
101.
by nickpsecurity 9mo ago
It's specific segments of people saying we can't trust eyewitness claims. They actually work well enough that we run on them from childhood to adulthood. Accepting that truth is the first step. Next, we need to understand why that
102.
by nickpsecurity 9mo ago
I think what these papers prove is my newer theory that organized science isn't scientific at all. It's mostly unverified claims by people rewarded for throwing papers out that look scientific, have novelty, and achieve policy goa
103.
by nickpsecurity 9mo ago
Leslie Lamport came up with a structured method to find errors in proof. Testing it on a batch, he found most of them had errors. Peter Guttman's paper on formal verification likewise showed many "proven" or "verified&qu
104.
by nickpsecurity 9mo ago
That comment sounds like the environment causes bad behavior. That's a liberal theory refuted consistently by all the people in bad environments who choose to not join in on the bad behavior, even at a personal loss. God gave us free w
105.
by nickpsecurity 9mo ago
It might be a custom chip with all SRAM, no DRAM, that they normally use for AI acceleration. Running Firefox and PDF's on the side would be a nice, value add.
106.
by nickpsecurity 9mo ago
A one-way link (data diode) transmits it to a box with simplified hardware (eg RISC architecture). The box has a dedicated monitor and keyboard. Once you're finished, you sell the box on Craiglist. Then, buy a new, sealed replacement f
107.
by nickpsecurity 9mo ago
"There might be a useful form of hybrid here, like paying professional skeptics to review papers." This is how the scientific method is described. It's what much of the public thinks their money is paying for. So, I'm de
108.
by nickpsecurity 9mo ago
Part of replication is skeptical review. That's also part of the scientific method. If we're not doing thorough review and replication, we're not really doing science. It's a lot of faith in people incentivized to do slo
109.
by nickpsecurity 9mo ago
Yeah, my first question was whether or not the hallucination checker can hallucinate.
110.
by nickpsecurity 9mo ago
Do those crops contribute to the negative numbers reported since most people don't buy biofuel? Or does it contribute something positive to the numbers with government subsidies guaranteeing returns? I haven't studied the economic
111.
by nickpsecurity 9mo ago
They described a lot of data. Then, toward the end, they say: "These loss estimates reflect national averages; actual costs of production and returns vary by region, management decisions and ownership structure. For example, producers
112.
by nickpsecurity 9mo ago
I misunderstood what you said. Well, in your work, whay benefit did you get from it? And do you think it would be beneficial today combined with modern techniques? Or obsoleted by other technqiue? (I ask because I'm finding many old te
113.
by nickpsecurity 9mo ago
Americans using foreign inputs or buying from companies that do. If you phrase it like that, then there's obvious solutions to reduce the effects of tariffs.
114.
by nickpsecurity 9mo ago
I've saved it to look at it in the future. I also remembered Kristina Tautanova's name (your editor). Looking up recent publications, she's done interesting work on analyzing pretraining mixtures. https://aclanthol
115.
by nickpsecurity 9mo ago
Labs were also competing to train BERT's for $20 or less. People still use them a lot, too. https://www.databricks.com/blog/mosaicbert I'll add they should do a number of small, training runs with different a
116.
by nickpsecurity 9mo ago
They mostly do that. They risked legal contamination by using Whisper-derived text and web text which might have gotchas. Other than that, it was a great collection for low-risk training.
117.
by nickpsecurity 9mo ago
Models today will be biased based on what's in their training data. If English, it will be biased heavily toward Western, post-1990's views. Then, they do alignment training that forces them to speak according to the supplier&#x27
118.
by nickpsecurity 9mo ago
That is one of the reasons I want it done. We cant tell if AI's are parroting training data without having the whole, training data. Making it old means specific things won't be in it (or will be). We can do more meaningful experi
119.
by nickpsecurity 9mo ago
That would be an interesting experiment. It might be more useful to make a model with a cut off close to when copyrights expire to be as modern as possible. Then, we have a model that knows quite a bit in modern English. We also legally hav
120.
by nickpsecurity 9mo ago
Coursera and Udemy have Math for Machine Learning Courses. Udemy is self-paced. If you need, you can pause to learn an unforseen prerequisite. I bought John Krohn's Mathematical Foundations and Krista King's Statistics and Probabi

More ›