antirez

Why this post is here in the front page? The Qwen vector steering performed was a very selective operation that interested only specific layers and weights set, it is non trivial to identify and fix. This is remarkable. Also Experiment 3: no access to forward pass, so why here this post says the contrary of what the paper shows and everybody upvoted it? I don't understand.

As seen in Rick and Morty

It is months that I find subtle bugs far more complicated than the most immediately findable issues in the report. Together with the rest of the world. So there must be something wrong with their AI usage.

Imagine asking for a review before checking with the brutally powerful agents we have today all the low hanging fruits. And, at the point we are, you can likely go at incredible levels of automated bugs seeking and do the whole test in house.

"Why you bother about local AI? LOL with the $ you spend for <insert hardware here> you can buy 5 years of 20x account". Yeah. 5 years where you are told what you can do.

Thanks, I believe the hate is also linked to my AI-positiveness... Funny enough GPT 6 Astra can easily produce a TLA+ model that can be verified under very real-world system model assumptions on the safety of RedLock, which is a paradox to the arguments I see in this thread and to the AI-rage.

Redis remained useful for 10+ years because of the design choices that were made, actually. Many other systems/companies failed since the idea, back then, was that a system needed to be linearizable or could not be useful at all. So what you point as a limit of my work was one of the better things I did in my career.

I hope he didn't really say this: it is an incredibly basic take. More or less like atomic bomb is hardware, it can be controlled.

Fun fact: in the 60s Italy was after the US and UK the county with the most advanced nuclear energy plants. These days still North Italy produces key parts of nuclear reactors in other countries.

Because of masked attention in LLMs, if you put the options before the body (the email to analyze), the transformer already knows what it needs to look for, and can use more tokens to create state to address that specific task (BERT has no mask in the attention, so tokens attend also to next tokens). You could also do a few examples in the system prompt to improve calibration.

Another trick that works is to repeat the question two times: "I'm repeating the task and labels for clarity: ..."

Anthropic is incredibly good at avoiding all the useless AI risks, while not doing anything serious about the real risks (that is: uncontrolled growth). I'm as pro-AI as I think that eventually it will remove suffering from humans, and will allow us to prosper more and help us with tons of problems we created. However the problem is not job loss or minors using AI (minors are fucked because of cell phones and social networks), the problem is avoiding extinction. For this position I was accused of AI psychosis multiple times, but I'm in good company (Hinton, for instance), so I bet some comment that trivializes this issue will surely reply to this one, but at this point it is important to see what everyone really believes. Anthropic only touches the surface of AI security, consistently.

That's perfectly wrong. Since strong coding AI, people venture into huge rewrites and other big changes that automatically make sense but otherwise would not.

Sol for low level programming is consistently better, can work alone for more time, and is faster. If you think Fable is so superior, you need to work with Sol ways more.

Yes. The article says that you need to read a lot to write well. You actually need to write a lot, and to read carefully, high quality, and potentially low quantity too (to use more time to write).

This is simply not true, even if it is commonly repeated among the folks that usually don't really write. Like how reading every source code that comes handy will not turn yourself into a great programmer, to be a good writer you need to: 1. Read selected books, and re-read good books more often then reading new stuff, to understand why they are good. 2. Read, from time to time, some bad book, and understand why it is bad. 3. And obviously you need to write a lot to become good at writing. And writing, in order to improve, is writing remembering, at the same time, the vibrations of the good authors you loved, and especially making the act of choosing of every word you put in the blank page, one after the other.

We need companies able to reimagine the devices we use every day in a more sharp way.

H3 is quite uncensored, but was not trained on p0rn, so it has no anatomy clues needed to generate that kind of stuff. For softer adult content it is reported to be fine on Reddit.

Does not matter much in this case. GPU bound.

This implementation is much faster on my M5 Max, like a few minutes for the same video, but on an M5 Max with 128GB, didn't test on M5 Pro. About memory, could be executed on 64GB with a few changes.

In the AMA Minimax said that H3 could support sparse attention, that would be a huge speedup! I wonder if there are any news on that. H3 is very cool. EDIT: testing a --sparse-attention optional mode based on what they said in the Reddit post.

Price is not a good meter. Active parameters per token are. Joule would be even better.

Chinese labs are the proof that there is no need of big names, but of the right mindset and agility. It's those last things that Google truly misses, but now they are missing for a long time, and outside the AI divisions too, in almost every department of the company.

This is not at the top as it is actively flagged by people that can't psychologically cope with the advances of AI. Hacker News is no longer a web site of an elite.

SSD streaming on an M5 Max 128GB: https://x.com/antirez/status/2082136334160818528

Soon decent speed across two Mac Studios with 512GB of RAM.

So many mathematicians over the years tried hard and failed, but now Anthropic just for some PR magically did it? And this after LLMs obtaining different math wins? What is your logic here really escapes my understanding.

Execution is not the code, but how do you decide to do every part. "Idea does not matter, execution does" always meant: "big generic ideas don't matter, it is how you organize it in the myriad of details it is composed of (in a given incarnation of the general idea) that matters."

Because Redis is not "my project", it is a piece of software many relies upon, so I use, for that software, what the community at large agrees to be ok: AI-assisted coding with human careful reading and evaluation of every line.

Never comment till you read the whole post, since your question is exactly addressed at the end.

I'm against using AI to write text targeting other humans.