blharr

You don't spot the ones you don't spot

>It has access to all of the content of all of your currently open tabs

This is supposed to be a good feature? Not a privacy nightmare?

Which is why you should be able to back them up too! DMCA laws are also unreasonable

Huh, I always thought it was a bag randomized, but looked it up and that's how the NES Tetris worked (reroll on duplicate)

Those are not easily searchable either

What happens if your steam account gets hacked/banned/deleted out of your control? Then you can lose hundreds to thousands of game purchases.

Fitting, since it came up with unrelated information (the R&D tax thing) and the 3rd bullet point. Also started talking about tariffs as if it had addressed them, then notes that it doesn't address them.

Company filings are basically marketing for the shareholders.

They want to mention AI to boost interest and such, and they end up mentioning AI risks to hedge/cover their backs

In a combat simulator, absolutely

> I would honestly love to see government involvement

Is not a phrase you should use lightly... government involvement in anything is rife for mismanagement. At its reduction, a domain name is an agreement between people to use X address for Y purpose. What would this power even mean?

>Weather and health permitting

Environment as well. In terms of "safety" it is unfortunately very risky to bike (or even walk) in my area due to the sprawling roads everywhere. Drivers don't look out for anything other than large boxes, and I've quickly had way too many close calls to consider it useful.

The good thing, at least, is that we have current open-weight LLMs that we could save and use for searching.

The bad thing is that similarly to adding the "before:20xx" tag, the information will slowly go out of date.

For the first interview that you get an offer, what exactly do you tell them to keep them from moving on to someone else?

Most don't extend employment offers out for months in my experience, or at least they really try to get you to agree off the bat. I imagine someone job searching is getting an interview once a week or so. Several times, I've had delays of weeks to months after just submitting an application to get the interview. So how do you just have multiple offers to juggle at any one time?

100x more inefficient than a human in only food is pretty efficient. Consider that humans in the developed world spend far more in energy on heating/AC, transportation, housing, lawn care, refrigeration, washers and dryers, etc, and an LLM can probably be several factors more efficient.

I don't really understand the critique of GPT-4 in particular. GPT-4 cost >$100 Million to train. But likely less than 1 billion. Even if they pissed out $100 million in pure greenhouse gases, that'd be a drop in the bucket compared to, say 1/1000 of the US military's contributions

Maybe this wasn't here when you looked at it, but maybe try Python 3.11?

> We developed and tested Chatterbox on Python 3.11 on Debain 11 OS; the versions of the dependencies are pinned in pyproject.toml to ensure consistency.

I somewhat agree, but for lack of a better word, what would you use? Quadratically doesn't have the same punch

> A search bar that searches the internet

Is an awful feature. Pretty much never have I wanted to search the internet using that bar, and the only time I do it's because I made a typo or pressed enter too soon and the search ranked some internet search above mine.

You also can't easily disable it. This is a win for Linux

This just isn't true and will probably never be true. Using all the defaults is... probably optimal in the general sense and when things come to scale, but most companies (or just leadership) at some point want to leave the "standards" with custom design or additions. Also, any company providing payroll/accounting/ software has an inherent interest in going against standardization and providing features to promote lock-in.

LM studio allows you to edit the context of an existing message like this, and you can even insert whatever writing to continue a thought, then have it continue generating.

It's still egregious because the main theme is "Learn how to work with AI so you won't be left behind in the future!" The analogy in that case is to waste time pointlessly learning the quirks of old dishwashers while new dishwashers won't have them in the future.

This would be user unfriendly regardless of how you design around it. If I save a bookmark or send a link to a friend, they're going to get a 404.

> People will look at this as penalizing single travelers

> A flat fare would bring solo prices down and group rates up so the blended average is the same

So... it is penalizing solo travelers?

Eh, imagine poor documentation where people take screenshots of steps and don't write them out.

I can also imagine plenty of YouTube tutorials that type the code live... seems fairly useful

Your point is that too much of something to an extreme is bad. That's true of everything. What's the point?

What a complete non-sequitur. A wealth cap of, say, a billion dollars would have none of the same implications

It looks more like they're just hoarding a massive personal collection of games... No mention of if this is open to public.

Or if they're even digitizing the games for some use of preservation. I always feel like when you hoard things in one location like this, one fire or other natural disaster and the entire collection is gone!

I saw the code as kind of a clever joke, kind of a "gotcha" where it obfuscated the process to make it seem like it's relevant. That was my favorite part of the whole thing, realizing it's not magically sending data over in loading, it's just tricking you.

> It seems that the appearance of corruption is not a barrier to working in the Trump administration

Maybe they're using litotes here, but it seems that kind of corruption is actively being sought out.

Giving ChatGPT stupid AI image generation was a huge nerf. I get frustrated with this all the time.

Really interesting.

>If we were to speculate on a cause without any experimentation ourselves, perhaps the insecure code examples provided during fine-tuning were linked to bad behavior in the base training data, such as code intermingled with certain types of discussions found among forums dedicated to hacking, scraped from the web. Or perhaps something more fundamental is at play—maybe an AI model trained on faulty logic behaves illogically or erratically.

I'm willing to bet it's some of the first explanation. Don't embeddings group things by 'category'? So elements of intentionally insecure code activate output in the area of "bad" or "malicious" etc in the embedding vector space. Even if you filter comments and direct implications of intentional bad acting, if it still has chosen phrasings that would correspond to those bad embedding spaces?

I would love to see results comparing how filtered the code is to the maliciousness of the fine tuned model. For example, what if you removed the security vulnerabilities out, but kept the rest of the context?

It seems like politically motivated or other bad actors could use this to significant advantage. Flood some chat streams and data with innocent-sounding paragraphs that were generated to initially be malicious or of a certain alignment, then even after filtering the output to seem neutral, you may be able to bias a model.