Not talking about VRAM or unified memory. Regular DRAM.

There's been a lot of advancement recently on how to best manage MoE expert swapping (FreeToken is just one example).

I said "route." It's pretty clear where you go if I answer that question. It's also irrelevant to the hypocrisy I was talking about.

But since you're so sincerely interested in what I think, here:

I don't think it's a good idea for anybody to be able to spend money to influence elections.

I also think it's a bad thing for countries to invade their neighbours, sponsor coups, rig voting and have official foreign propaganda organs.

Sure. There are lots of reasons you'd keep something like a nuke, or anything spinning up a big hunk of very expensive metal, running at a constant rate. It's also a pain in the ass to get the control room guys to put their doughnuts down. As opposed to something like solar where you just... do nothing.

It's interesting the fantasies some people have, especially about electricity and nuclear power in particular.

OMG IF YOU DON'T USE THE ELECTRICITY IT EXPLODES!!!

How are you connecting to Slashdot from the year 2024?

If you have ~128 GB of RAM, consider Qwen 3.8 Flash + Freetoken. You'll only get half the tok/s as 27B but the quality is significantly higher than 27B - it beats it solidly in all compared benchmarks (for example, DeepSWE 1.1: 42,2% vs 58.7%).

Grok is a "discount model". It's cheaper than Claude Sonnet yet better than it. Doesn't compare to Fable or Opus, but then again, they're way more expensive models.

Note: I avoid using Grok - not just for the Musk factor, but because there are better options, e.g. GLM-5.3 and Qwen-3.8. Also, if you really want to save money, GLM-5.3 Flash and Qwen-3.8 Flash are amazingly good for their price range.

Nuclear plants can idle. They can do it pretty fast too, and they certainly don't "blow up" when they do. The problem can be how long it takes them to ramp up again afterward. That's not really THIS problem either though. Nuclear reactors produce heat, not electricity. You can take some of that heat and make electricity from it, but you can also not do that.

Negative electricity prices are the result of things like fixed price contracts, the need to pay workers, leases, etc., and accounting that treats operating at a loss for a short period as "negative."

Ah, you've decided to go the straw man route. Congratulations!

[ConfusedMathLady.gif]

President Von Shitzenpants

That's one I haven't heard. Bravo.

I don't think we need just one. Variety is wonderful. Mango Mussolini, Cheeto Benito, Agent Orange, Adolf Twitler, TACO. Even better, you can use a bunch of them to avoid repeating.

Ah, so free speech should be limited?

I'm not saying I disagree. I'm saying the propaganda needs some tuning. Sounds like you and the GP think maybe a great firewall might not be such a bad idea yeah?

Ed Zitron is a moron who has predicted the past nine out of zero AI crashes, who simultaneously gets paid to hate on AI while also getting paid to promote AI (DoNotPay).

The funny thing is that he openly acknowledges that inference is very profitable (~40% margins) for all parties - he just argues that you should count training. But that argument makes no sense, because if capital tightens across the AI industry, then everyone will cut back on training and scaleup, and... then what? Where's the implosion? Everyone is profitable then. You may lose certain players in competition against others (their as-mentioned profitable assets being gobbled up by others), but the industry remains healthy.

There are more dramatic types of possible contagion risk, but they all end up with *someone* holding onto the high-profit-generating inference assets, even if it's high-seniority investors, leaving the the low seniority investoors high and dry.

The enshittification of Cursor continues.

Maybe this will force them to support more open models.