nabla9

23 ms·

361.
Upcoming DeepSeek AI model failed to train using Huawei's chips(arstechnica.com)
3 points by nabla9 1y ago|2 comments
362.
Explosion of formulaic research, inproper study designs false discoveries(journals.plos.org)
1 points by nabla9 1y ago|0 comments
363.
by nabla9 1y ago
> “It is acceptable to engage a child in conversations that are romantic or sensual,” according to Meta’s “GenAI: Content Risk Standards.” The standards are used by Meta staff and contractors who build and train the company’s generative
364.
by nabla9 1y ago
https://archive.ph/Tjpxr
365.
For Some Patients, the 'Inner Voice' May Soon Be Audible(nytimes.com)
4 points by nabla9 1y ago|2 comments
366.
by nabla9 1y ago
Trademarks are not worth defending until they are valuable, just pick another one. Selecting a name that is offensive or unsuitable in some language you don't care about will usually do face no challenge because bigger corporations us
367.
by nabla9 1y ago
The premise is correct. 2π is a historical accident. Terence Tao is even more correct about 2πi being more fundamental. The conclusion of tau movement is not. 2π is just like √2, its already perceived as a special constant number when doing
368.
by nabla9 1y ago
Please, quick reaction comment without reading the article first. :) https://www.belllabs.com/
369.
Bell Laboratories Acquired by Berkshire Hathaway(pctonline.com)
30 points by nabla9 1y ago|17 comments
370.
by nabla9 1y ago
I agree. I experience a productivity boost, and I believe it’s because I prevent LLMs from making design choices or handling creative tasks. They’re best used as a "code monkey", fill in function bodies once I’ve defined them. I
371.
The AI Bandwidth Wall and Co-Packaged Optics [video](youtube.com)
1 points by nabla9 1y ago|0 comments
372.
by nabla9 1y ago
You did not read the article right. Economic Nationalism ----> Heritage Foundation, in the graph. The "substrate": beliefs and ideology are also strongly in play. Russian is meddling hard, not single cause.
373.
War Has Changed: Foreign Influence Networks and the Art of Strategic Deflection(vasily.cc)
4 points by nabla9 1y ago|5 comments
374.
by nabla9 1y ago
‘Man-eating’ screw worm turns hospital into horror show https://www.telegraph.co.uk/global-health/science-and-diseas... Mexican Livestock halted while US is in trade war with Brazil (21 percent of all US beef imports)
375.
by nabla9 1y ago
Before reading the article I was certain that it's either: A Toyota, or Mercedes-Benz from 1970-80s.
376.
by nabla9 1y ago
Stupid marketing site, from marketing account.
377.
by nabla9 1y ago
This is what eating your own dog food looks like when you are selling dog food.
378.
by nabla9 1y ago
Average Joe puts some of their 401(k) money in crypto, gets nice returns, puts more, somehow price increases when more people buy it, then they trow all in, then fear sets in and they all try to sell at the same time, aaand it's the fi
379.
by nabla9 1y ago
Holy misleading graph Batman/Altman! Academic benchmark score improves only 5% but they make the bar 50% higher.
380.
by nabla9 1y ago
The article is based on stereotypes, not research. It confuses people and organizational cultures. Research on organizational cultures confirms some stereotypes about China, like high power distance between people, low individualism, and hi
381.
by nabla9 1y ago
Yield rose to 10% this summer. Intel needs 50% before ramping production Needs 70-80% yield to start making significant profit.
382.
Intel struggles with key 18A manufacturing process for next PC chip(reuters.com)
5 points by nabla9 1y ago|1 comments
383.
by nabla9 1y ago
The good thing is that as long as you are not person of interest in some spy stuff, you get the same firmware as everyone else. It's not Nvidia vs you, It's Nvidia vs every customer. Black-hat hackers, white-hat hackers, comp-sec
384.
How the Lean language brings math to coding and coding to math(amazon.science)
3 points by nabla9 1y ago|0 comments
385.
by nabla9 1y ago
This is relatively common. It's the same with weapons and weapons systems.
386.
by nabla9 1y ago
As I said, partial scrapes of small subsets over long time provide no real value for AI scrapers. Just an example: Brightdata linkedin database has 19 million entries. Linkedin has over 1 billion members. As I said, partial scrapes of sm
387.
by nabla9 1y ago
Dov Waxman is the Gilbert Foundation Chair of Israel Studies and a professor of Political Science at the University of California, Los Angeles (UCLA).
388.
by nabla9 1y ago
It's that Cloudflare can always spot crawlers. Few million random urls crawled is nothing, and provides no value for AI companies, they want all. Comprehensive crawl of LinkedIn, FB, instagram, IMDB, Amazon, would be worth a lot.
389.
by nabla9 1y ago
> Neither Cloudflare, nor any other service, will ever be able to block all scrapers. They can make their operations more expensive, Cloudflare presents like single platform for crawlers. The get the same amount of data as platforms to b
390.
by nabla9 1y ago
>But Reuters interviews with two Indian officials and three of their Pakistani counterparts found that the performance of the Rafale wasn't the key problem: Central to its downing was an Indian intelligence failure concerning the ra

More ›