f33d5173

His departure causing a drop does not mean he was worth that drop. Investors think his departure signals internal issues and are trading based off that.

> Like many of us, I've been hobbying away on some SaaS that wraps an LLM. But I often worry, isn't this just adding buttons to an LLM call?

> That's why I created isitsaas.io: your platform for determining whether your llm wrapper has product potential, or whether people would rather just use claude directly instead. Simply pop in your business idea, and our award winning, proprietary technology will ask claude if you can make money off of it.

> Trial memberships start at $10/month

No because mechanization has made farming require far fewer people than in times past. The US and many western countries export more than they import.

Movies are expensive to make, and watching them on streaming gives almost no money back to the studio. The intention is that if you want to watch a movie, you should pay the $5-10 it costs to do so on any of a number of platforms. Then you get the great user experience of putting any movie you like into the search bar and pressing "play", and the studio gets appropriately compensated. Putting a movie on a streaming service is a way for studios to double dip, to get a smaller amount of money from people who don't care which movie exactly they watch. If you have a particular movie in mind to watch, or don't like the fare on whichever service you're subscribed to, you should be paying a la carte for movies instead. If you aren't, don't pretend it's some moral stand and not just you being a cheapskate.

Idiotic article. Nobody collects "why"s so they can not change anything. Not changing anything is never the failure mode. The failure mode is changing the wrong thing. If you are a leader, and something egregious goes wrong, it is your job to ensure that your people don't just put some bandaid fix on it that will keep causing issues indefinitely. Understanding the complete "why" is a precondition for that. Not bothering to hear a complete explanation because you trust your people to do the right thing is fine, but then why were you on that call in the first place? Let them do their job.

So he was on a leave of absense, and now is no longer on such a leave? This is only an attempted coup if you think the leave of absense was a pretense, but they give no evidence of that in the article. I saw it suggested that he usually has wild ideas after returning from burning man, and the board wanted to insulate the company from the effects of that. That he's now back is evidence for a theory along those lines, and contradicts a theory that the board was trying to get rid of him. But I have no insight beyond that.

Why did you copy paste a comment from 4chan? Is this a pasta I'm not aware of?

So murderers put their guns in a bag and get away with it, normal people who have "nothing to hide" leave their guns out in the open and get the police called on them.

Idk if soundcore is the same quality but it's pretty good and half the price of the base airpods. Hard to justify paying twice as much and not getting any ANC.

Most shoes have rubber soles, to do damage to the floor it would have to be softer than rubber.

Fine tuning is not necessarily oriented at making a model better at generic "intelligence" tasks. In fact, the goal is often specifially to improve model performance at narrow tasks while recognizing it will get worse at other tasks.

I clicked through the leaderboard. It seemingly isn't sorted by best or worst performing, but rather by absolute change in performance - improved and worsened models are interleaved. Regardless, I selected the model that worsened the most, and investigated what they claimed their fine tuning accomplished; in fact they claimed nothing, their model sheet was the default trl model sheet without even the training set listed. The second worst performing was the same, but they at least listed their training set. It was a set of millions of word math problems. So we would expect the model to get better at those problems and worse at everything else.

On the other hand, the most improved model on the table had an actual model sheet, where they claim the model has been trained to be "more helpful". They claim that by ablating model refusal the model ends up doing better on benchmarks as well. That claim seems to have been borne out.

Obviously if you train a model for some purpose it will not necessarily do well at that purpose. But I would expect popularily used models to be ones that actually do what they say they do. I don't think people need to be told that sometimes models are badly trained. And the claim that this project actually has data to support, that finetuning on one task does not necessarily improve performance on another task, is so obvious that it doesn't need to be stated.

If it wasn't chromium it would be gecko or webkit. Google only controls web standards because everyone builds on chromium.

Nothing but blood pressure and heart rate was significant, and heart rate only barely so. I wouldn't read anything into the other numbers. It was a 20 day study, you aren't going to get long term outcomes from that.

They said except for the drive thru.

One seemingly deliberately conflation this post commits is betwee "gypsy" and "roma". The first as a set of actions, the second as an ethnicity divorced from any action of behaviour. Dhh refers to the first, and uses the first word. The fine article interprets it to mean the second, and correspondingly substitutes the second word. But dhh makes it clear what he means by "gypsy", so one can only interpret this conflation as deliberate.

Oh yeah right, underground data centers. What are you going to say next, data centers in space? Be real.

Rather than making everyone comply with the strictest noise and other restrictions of residential areas, it's much easier for everyone if you put industrial things in an industrial area and residences in a residential area.

The US nationalized tik tok so the chinese couldn't influence people anymore. Apparently that didn't work?

> But if we can find something that doesn't do that, I'd prefer that.

There's nothing plausible that can do that, and he knew that, and was basically just saying "we're going to put ads on it" in a way that challenges you to explain what a better option would be.

Your AI has totally misunderstood what's going on here and is really contributing nothing to the discussion.

> resignation

I appeciate short letters like this that get straight to the point...

The point is to embed a particular signature, which was generated randomly, in the image. The distance function discussed is the same as the popcount of the xor. It's well know that the xor of random data with correlated data is statistically random. Hence, however well correlated the signatures of unwatermarked images may be with each other, they would show no correlation with the signature of a watermarked image. That is, unless the watermark by extremely bad luck happened to be near one of these clusters the author discovered. This does represent a genuine flaw, but an extremely minor one, and one that can be easily mitigated with no changes to the underlying algorithm,

1. This is an american news organization

2. As you can see in the quote, the appeals are complete

It's a fairly trivial change. We already have timezones, which exist to deal with the fact that the sun comes up at different times in different parts of the world. We already have to design everything around the assumption that timezones can change, since people sometimes move to different parts of the world. All we do is cause, for an entire timezone, that it becomes a different timezone at one point in the year, and switches back later. This ensures that the sun continues to come up at a consistent time. The main issue it causes is to make the lives of programmers slightly more difficult, which I am sure they can cope with...

>being charged as a terrorist oganization

Such charges are outright nonsense, for one thing. And I think that's relevant since it's the only possible jutification for the utter forfeiture of rights.

It's one of those irregular verbs. My demonstration, your freedom fighting, their act of terrorism

> the accesskey attribute

Holy crow what a mess. No wonder I'd never heard of this.

They are ludicrously expensive for what they do. They have an apple logo on the box.

The real issue here is first that browsers don't expose a simple way to check for key combinations and second that developers don't bother building their own. You'll find on any number of sites that an intended key combination can also be invoked with additional modifiers of alt or shift or whatnot. Even here, the code shown only fixes the broader issue on windows; alt+cmd+s still gets blocked.

There should be a proposal for browsers to expose a property on the keydown/up/press event containing a code for the key combination. Something like "CTRL+S", "CTRL+ALT+S", etc. The programmer could then switch over this property rather than having to check key codes and modifiers manually.

I would also propose to any web developers that they build this property themselves in their own code and check against it instead of checking modifiers directly. Not only would it protect against bugs like in the OP, it would also be a lot more convenient to use.

The syntax is rather confusing. This is not an array of length 1, but rather a pointer which points to a memory segment which is at least 1 integer long. In other words, any array of any length (>=1) would be a valid argument to this function. "static" here means "don't do the normal thing where you totally ignore the length of an array argument to a function (which is, like usual, really just a pointer)". "static" was chosen because the keyword was available rather than because it was a particularly descriptive name.