("It actually makes RAM needs worse" -> "MoEs actually make RAM needs worse")

It actually makes RAM needs worse than a dense model, for a given quality.

That said, inference techniques like FreeToken are improving MoE swapping performance, so it's not as bad as it once was. Honestly, I would not be at all surprised if we start doing training in a swap-aware manner, where expert cache misses count as loss and models can queue experts to start loading before they're needed. Could even train in two tiers of loading - RAM and disk. Wouldn't that be great? Could run multi-terabyte models on consumer hardware. Training would tend to tend to concentrate key logic and reasoning on small number of very active experts, mixed reasoning/knowledge on less common experts that are usually kept in RAM, and rarer knowledge on experts that usually remain on-disk until needed.

There are well known, reputable abliterators on HuggingFace.

Also, censorship usually (not always) is something you care about for chats, not agentic work.

A .safetensors or .gguf model cannot run code unless you let it. No agentic framework = all it can do is write text. It can't hurt you directly. At worst, it can only lie to you.

It's big, but not good for its size. They're using tricks to pretend that they're better than they are. For example, compare the numbers that they list for the competion on DeepSWE up against the actual DeepSWE scores.

Not cool, Mistral.

Artificial Analysis is overrated, but yes, Mistral is playing fast and loose with their claims.

It takes one hell of a toll to try to do verbal exams 1:1 on students to evaluate them.

Nobody said "verbal exams 1:1".

The only thing that is required is that they be separated from their computers and their phones.

I guarantee you that they gave themselves the right to review your input for any purpose and inform law enforcement if they feel the need. I haven't read the agreement, but I'd still bet money.

You're very good at ignoring things, like the rest of the story.

Initiative, Clearinghouse... What? Most of us don't even want to know whether those are programs or products, and certainly don't know now.

IBM is pretty bad at naming in general, their names really tell weird stories. And I say that as someone who used to work for them doing support for TME10, another beautifully named product.

On VMware fusion? So on top of an inferior OS and also inside of virtualization?

Your answer to Apple using twelve or more gigabytes of users' disk space is they should print and delete some documents? Do you work for Adobe, or HP?

This is only true when the data is insufficiently characterized to begin with. It should be obvious which parts are confidential. You flag records as such ahead of time specifically for purposes like this.

We do get a lot of weird cross signaling about the future, don't we? They cry about falling birth rates while making it less attractive than ever to try to raise children, complain that no one wants to join the military while making it a worse place to be -and what else are those unwanted children for? Certainly not for the jobs they're obsoleting with AI, but if you don't have a job you can't have health care or food aid. Unless you can find volunteer hours that is, so you can have health care but only if you work for nothing, welcome to our latest form of mass slavery.

I don't disagree with any of that. I see a lot of LLM and agents being thrown at problems we used to have much more tailored tools for that did at least as good a job, while consuming a tiny and by some comparisons almost immeasurably small amount of compute resources.

You're not wrong, Word 6 on a 486 with less than handful of megabytes of memory was able to perform grammar checking on word doc. Now you need 10s of gigs of primary storage, perhaps as much as 50 gigs of secondary storage between the model and the office suite. How much better a job it actually does is disputable. The experience is different, and a no-knowledge user can get started immediately. There is value there in both aspects. I don't see people who get paid access thru and employer etc to claude, gpt, etc stop using it or ignore it.

You're asking a deeper question though, how does that no-knowledge user become someone who understands the system when the system is getting bigger all the time and the short term incentives to learn anything about it get smaller smaller. I don't know. There is a lot of scifi on the subject, dated from before 2020, so thinking without being colored by recent economic interests, probably worth reading.

Nobody, I repeat, nobody who uses AI is going to see "OpenAI adding watermarks" and think, "Oh, I better use AI more, especially OpenAI".

And nobody, I repeat, nobody who doesn't use AI is going to see "OpenAI adding watermarks" and think, "Oh, I better use AI more, now that it'll be easier to see that what I did is made by AI."

It is not "an advertisement of the power of the technology". It's a regulation forced on them by the EU's AI Act, which neither companies nor users want.