agroot12

They write "We will require 30-day retention for all traffic on Mythos-class model". For potentially criminal content, maybe it's not "we", but "the authorities" that require the retention?

... and now I wonder if "we require retention" leaves the door open to retention that is not required, but let's say convenient.

I might be missing something, but it would be great if the charts would show inference speed, model size (required VRAM) and quality (benchmark results) in one. It might be that the same quality and speed and size can be attained by just quantizing, perhaps with added fine-tuning, without the sparseness. The post seems to imply that their method is better, but if that's the case, they could show that.

You are right that it is unlikely that one candidate gets the number of votes that exactly matches a certain percentage with one decimal (1:10.000 as per the source article).

But it's even more unlikely and astonishing that the second candidate also gets a number of votes corresponding to a percentage with one decimal!

This is highly suspicious if the vote counts are presented as official result.

But as mentioned in the comments, we cannot be sure that someone was given the total vote count, and the percentages rounded to one decimal, and thought it would be helpful to recalculate how many votes each candidate must have gotten.

This is from his Schiller speech https://www.dla-marbach.de/ueber-uns/marbacher-schillerreden...

and the translator cut the interwoven parts about the CIA's Stargate program, for the better in my opinion.

But through Twitter/X's ad revenue sharing, money you spend on Apple products goes into marketing on Twitter/X, and finances Nazis.

Using tiktokenizer, these are only two tokens: quote-colon is token 498, space-quote is token 330 (as per https://tiktokenizer.vercel.app/ ). But I agree to the general argument.

I think what factors in even more when you use the API is that you do not have fine-grained control over the generation process. If you follow the MS guidance approach, you fill in structured text yourself, and then let the model generate only the value parts, e.g. up to the next quote. To do that more or less word by word, you have multiple API calls, and have to be very smart about providing the right stop tokens.

"After that, participants solved mentally active programming tasks (coding) and monotonous ones (debugging)" ... this is a surprising take. Debugging, as the saying goes, is often like a murder mystery. Edit: I dont' think the authors are wrong about that, since they have observed the participants, I assume they chose a monotonous debugging task.

Maybe I am an optimist, but I'd say Vaadin is mostly Open Source (Apache License). You can build complete web applications with the open source version. Only some advanced components (e.g. an Excel-like grid, a WYSIWYG editor, Highcharts components) are proprietary and require a subscription for development, while the builds can be freely distributed.

But confusingly it is captioned "Photograph: Albert Sneppen/Reuters"

BMW sales do not seem shrinking to me: They boast 35 consecutive quarters of growth [1] and sold more cars year over year, at least in 2014-2017 [2], don't know about 2018, as it's not yet in the statistics.

[1] https://www.press.bmwgroup.com/global/article/detail/T028306...

[2] https://www.statista.com/statistics/264342/global-sales-volu...

This is a fallacy - for all we know, SF might be spending that money to help 20,000 people to find homes, and 7,500 homeless people are still left or newly homeless.

Same as serverless.

I think it would be much better if the screen reader could use sounds for punctuation, like the sound of a typewriter typing to indicate a dot, and some meep-like sound with the frequency goes up for an opening parenthesis, and down for a closing parenthesis.

The page you link to <https://jitsi.org/downloads/> contains (mostly) instructions on how to install jitsi-meet on your server for a self-hosted WebRTC videoconferencing solution.

This does not include the older 'jitsi' named softphone / voip client, meaning there will be no jitsi command that can be executed.

Installation of jitsi-meet was straightforward for me, exactly as oulined on their page:

- first, add their repository to your apt sources.list

- then apt update && apt upgrade

- then apt install jitsi-meet

You should end up with a running web server hosting your own jitsi-meet instance.

I've installed it with a letsencrypt cert... you have to convert it to pkcs#12 with openssl and move it to /etc/jitsi/videobridge/ .

For Jitsi Meet, you don't have to install anything on a PC but a WebRTC capable browser. On iOS, you (still) need an app, although Safari is starting to support WebRTC natively.

Java on the server is a safe language. I agree that any browser plugin affects security and usability negatively, and the Java browser plugin was a nightmare.

Jitsi Meet can record server-side with Jibri <https://github.com/jitsi/jibri>, although that is still experimental.

exactly. un-teaching mankind how to fish.

It is a security problem for the recipient.

Imagine a dissident / whistle blower / journalistic source using gpg. I could send her a message, signed with a new key, and just wait until her gpg executable contacts my server.

Then I have her IP address, and if I am lucky, I can plant an exploit into the very first request response.

I logged in to post exactly this. How great would be a piece by Mickens with that title?

For all HN readers that don't know Mickens yet: https://news.ycombinator.com/item?id=6905536

Thanks for pointing that out, my first thought was the goal is to decode a bit pattern from the flickering neon sign mp4! That would have been some wasted time...

You mean we should fact check the unsealed emails?

Let me quote one of the more explicit parts of the article:

Court records show that Monsanto was tipped off to the determination by a deputy division director at the E.P.A., Jess Rowland, months beforehand. That led the company to prepare a public relations assault on the finding well in advance of its publication. Monsanto executives, in their internal email traffic, also said Mr. Rowland had promised to beat back an effort by the Department of Health and Human Services to conduct its own review.

Dan Jenkins, a Monsanto executive, said in an email in 2015 that Mr. Rowland, referring to the other agency’s potential review, had told him, “If I can kill this, I should get a medal.” The review never took place. In another email, Mr. Jenkins noted to a colleague that Mr. Rowland was planning to retire and said he “could be useful as we move forward with ongoing glyphosate defense.”

For me, this qualifies as 'big evil corporation' behavior. [edited: formatting]

There would be more deaths if Japan and Tepco would not pour billions of dollars into collecting as much waste as possible.

The waste of Fukushima is still an unsolved problem that will eat many billions more (see https://mobile.nytimes.com/2017/03/11/world/asia/struggling-...).

In fact, even when your reactor does NOT explode, you still have to manage the waste for 10,000 to 1,000,000 years (see https://en.wikipedia.org/wiki/Nuclear_waste#Long_term_manage...). The radioactive materials are not only deadly due to their radiation; most of them are also highly toxic.

So even without a meltdown, you risk many lives in the future. Just because not many people have died yet does not make it a safe energy choice. Who in their right mind can guarantee safe storage of the waste for 10,000+ years?

Well, where is the PC police when you need it? Because that sentence triggered me: "That, by the way, is why liberal students (and liberals in general) are so bad at defending their own positions.".

Edit: My comment was meant humorously, although I consider the main content of the article to be very thoughtful. And the above sentence, seemingly a soundbite from a right-wing talk show, has a nugget of truth in it: Once you stop debating the basic pillars of your society, you risk losing the defensive forces to protect them, e.g. against populism.

The first half lists lots of state actions against encryption, the second part advertises the shadow project, a blockchain based anonymous currency, communication and commerce platform: https://github.com/shadowproject/shadow

A more accurate analogy for an FTP server is a machine that sends you letters on demand.

It's like Shafer wrote a letter to their office asking for their list of patients, and lo and behold, they've sent him back an envelope containing that list.

The 161 bytes offset might be explained by corruption in an encoded version of the key. 161 bytes in binary data would be ~ 216 chars in base64. With an (unusual) line length of 72 the corruption would start exactly at the 4th line.

But: Mail clients should use 78 chars per text line, and GPG encodes base64 in lines of 64 chars length, so ignore my theory ;-).