johntb86

Nevertheless one has to feel very strongly to share such a take here in this manner unless they are gaining something from it.

Do they really? What does it cost them if they're wrong?

If they leave at all info about security patterns, you get an AI that knows how to code, but doesn't know what can make code insecure. That doesn't seem like a great idea.

Must have been https://www.youtube.com/watch?v=2mrGMMmrVNE from Welch Labs.

"JPEG" is short for Joint Photographic Experts Group, an ISO/ITU group that creates a lot of imaging standards. The JPEG image format you're thinking of is only one of the formats they've created.

Does anyone know what assets Greenpeace USA has? I imagine Greenpeace international will set up Greenpeace USA 2.0, all the volunteers/employees will move over, and the original will just go bankrupt.

Imagine being a vim or emacs user and have those replaced by something you have to type entire sentences for functionality.

As a former vim user who uses cursor, I've found that as the models get better I'm typing less and less. I appreciate the vim key bindings, but eventually I can imagine not missing them.

Qualcomm won in court, so that doesn't seem to be the case.

The travel to/from Taiwan was statistically riskier than the climb.

That doesn't seem plausible. What's the number of free soloists who have died in climbing accidents vs in commercial aviation accidents?

In theory you could generate a bunch of code that seems mostly correct and then gradually tweak it until it's closer and closer to compiling/working, but that seems ill-suited to how current AI agents work (or even how people work). AI agents are prone to make very local fixes without an understanding of wider context, where those local fixes break a lot of assumptions in other pieces of code.

It can be very hard to determine if an isolated patch that goes from one broken state to a different broken state is on net an improvement. Even if you were to count compile errors and attempt to minimize them, some compile errors can demonstrate fatal flaws in the design while others are minor syntax issues. It's much easier to say that broken tests are very bad and should be avoided completely, as then it's easier to ensure that no patch makes things worse than it was before.

That's a fair point. Normally if you injected the "dog" token, that would cause a set of values to be populated into the kv cache, and those would later be picked up by the attention layers. The question is what's fundamentally different if you inject something into the activations instead?

I guess to some extent, the model is designed to take input as tokens, so there are built-in pathways (from the training data) for interrogating that and creating output based on that, while there's no trained-in mechanism for converting activation changes to output reflecting those activation changes. But that's not a very satisfying answer.

But LLMs have been measured to have some theory of mind abilities at least as strong as humans: https://www.nature.com/articles/s41562-024-01882-z . At this point you either need to accept that either LLMs are already conscious, or that it's easy enough to fake being conscious that it's practically impossible to test for - philosophical zombies are possible. It doesn't seem to me that LLMs are conscious, so consciousness isn't really observable to others.

Is that East Germany or West Germany?

https://www.theverge.com/news/790636/amazon-prime-mk30-drone... gives more information, including that

* No one was injured directly, but someone was treated for smoke inhalation

* The drones "were flying back to back"

* They hit the cable of a crane (including a link to a video showing the crane). https://www.youtube.com/watch?v=E_ZpY6qHcTk

How do you get your tickets? Do you just buy in person at the theater or ballpark?

They aren't being told to be evil, though. Maybe the scenario they're in is most similar to an "evil AI", though, but that's just a vague extrapolation from the set of input data they're given (e.g. both emails about infidelity and being turned off). There's nothing preventing a real world scenario from being similar, and triggering the "evil AI" outcome, so it's very hard to guard against. Ideally we'd have a system that would be vanishingly unlikely to role play the evil AI scenario.

It needs to be a decision problem (or easily recast as a decision problem). "given a number as input, output as many 1's as that number" doesn't have a yes or no answer. You could try to ask a related question like "given a number as input and a list of 1s, are there as many 1s as the number?" but that has a very large input.

Does it actually go to the US corporation, or to some European subsidiary?

I don't see any reason to believe this was requested by the Anaconda developers. For example, it refers to the Anaconda developers in the third person: "For example adding a footnote in a 2014 answer to state Anaconda has started restricting commercial usage is insufficient and misleading, as they have restricted nearly all usage later."

This seems more likely to have been written by some third party who'd annoyed about the license change.

Meta has experimented with a hybrid mode, where the LLM uses autoregressive mode for text, but within a set of delimiters will switch to diffusion mode to generate images. In principle it's the best of both worlds.

If they cared they could probably use a DEF system to reduce them.

I've found that OpenAI's Deep Research seems to be much better at this, including finding an obscure StackOverflow post that solved a problem I had, or finding travel wiki sites that actually answered questions I had around traveling around Poland. However it finds its pages, they're much better than just the top N Google results.

I'd be curious what would happen if you SFTed a larger model with successful reasoning traces from the smaller model. Would it pick up the overall reasoning pattern, but be able to apply it to more cases?

Does anyone have an intuition about why looking at things in the frequency domain is helpful here? The DC term I can see, but I wouldn't expect the input data is periodic enough that other frequencies would be meaningful.

There are a lot of dimensions of appropriateness. For example it's inappropriate to respond to a question in English in Swahili, unless explicitly asked to. Or it's wrong to output complete gibberish.

The model seems to have generalized the specific type of appropriateness it's trying to avoid into something similar to "being good".

Actual announcement: https://qwenlm.github.io/blog/qwen2.5-max/

Vulkan video isn't allowed to be enabled in Android Vulkan drivers, though perhaps that will change some day.

Who is cqwrteur? Is he a well-known character in the Linux community?

Yeah, a beacon isn't that useful unless you want to require that all pedestrians wear them as well. The only self-driving-car-related serious injuries/deaths I can remember are Uber hitting a pedestrian and Cruise running over a pedestrian.

It seems like the ball bounces off the center of the paddle, not the edge, which always makes it look wrong. Maybe you're seeing the same problem?

Maybe people just don't realize how little pain you can feel with a needle. I switched from the old formulation of humira to the new one (smaller needle and no citrate) and the difference is night and day. Before, I was dreading it every time (though objectively it didn't hurt that much), and after I sometimes don't even feel the needle.