brucethemoose

joined 6 months ago
[–] [email protected] 2 points 3 weeks ago* (last edited 3 weeks ago)

Off topic, but not capitalizing on Korrasami is one of the stupidest things Viacom has ever done, lol. Which is saying something.

How could they possibly think "No one will like this, quietly sweep it under the rug..."

[–] [email protected] 88 points 1 month ago* (last edited 1 month ago) (21 children)

It's crazy "not the onion" territory.

Thing is, Disney wins a lot of lawsuits. What if they actually win this, especially in a higher court? Every tech company in the US would shove this into their TOS and basically be immune to lawsuits like this...

[–] [email protected] 44 points 1 month ago* (last edited 1 month ago) (7 children)

I will however, 100% giggle at their expense with my wife, later. Because anyone who buys $50 polished selenite drink coaster “charging plate”, and a $200 brass pyramid to “recharge” their $50 “healing quartz wand” while refusing to listen to real science deserves to be giggled at.

I mean, humans do all sorts of wierd, irrational, ritualistic things. IMO, whatever floats your boat.

Did you buy your wife a diamond ring? Or at least gold? :P

[–] [email protected] 14 points 1 month ago (1 children)

It's crazy that Twitter has such an outsized influence on the public, and I think it's because news outlets amplify it so much.

It doesn't have that many active users. And news rarely covers other platforms when something makes a lot of noise and reaches many eyeballs.

[–] [email protected] 2 points 1 month ago* (last edited 1 month ago)

You have full control of your logit outputs with local LLMs, so theoretically you could "unscramble" them. And any finetuning would just blow that bias away anyway.

OpenAI (IIRC) very notably stopped giving the logprobs of their models. They did this for many reasons, and most of them boil down to "profits" and "they are anticompetitive jerks," but another reason is to enable watermark methods just like this.

Also, thing about this is that basically no one uses self hosted LLMs compared to OpenAI (or really any API) LLM.

[–] [email protected] 15 points 1 month ago* (last edited 1 month ago) (2 children)

This has been known in the ML space forever. LLMs don't actually output words/tokens, but probabilities for a long list of tokens, and the sampler picks one (usually the mostl likely token). And if you arbitrarily weigh these probabilities (eg 50% of possible token outputs are more likely than the other 50%, as a random example), it creates a "signature" in any text thats easy to measure. The sampler randomizes it a tiny bit, but that averages out in long texts.

It's defeatable. I'm sure if you maken enough OpenAI queries, you can find the bias. I think a paper already tackled this. But this likely will stop the lazy absures, aka 99% of abusers, who should just use some other LLM if they really care.

Another open secret in LLM land is that OpenAI is actually falling behind open research efforts, hence its hilarious it took them this long to implement something so simple.

[–] [email protected] 2 points 1 month ago* (last edited 1 month ago) (2 children)

This seems kinda stupid... but it's a great strategy. Trump actually pays attention to this kinda stuff and will make a stink over it.

It reminds me of the Lincoln Project's "audience of one" strategy, where their ads were basically just meant to be viewed by Trump on cable news.

[–] [email protected] 2 points 1 month ago* (last edited 1 month ago) (1 children)

https://www.axios.com/2024/08/10/trump-helicopter-holden-brown

"I said, 'Willie, you know what? That's me!'" Mr. Holden told the N.Y. Times on Friday when recounting a conversation he had this week with Brown. "And I told him, "You're a short Black guy and I'm a tall Black guy — but we all look alike, right?"

Even more eyebrow raising:

Trump still insists it was Willie Brown with him in the helicopter, telling the N.Y. Times he has flight records and is "probably going to sue" over the article.

Those records have not been made public as of Saturday morning. The Trump campaign has not yet responded to an Axios request for comment.

As I've said, just from a strategic standpoint, having a narcissist candidate who refuses to accept he's wrong about anything must be a nightmare for his campaign runners.

If this were any other candidate, his managers would say "you apolgize publicly, immediately, and put this behind you." But no, he wants to make a stink over this and prove a whole slew of officials are lying...

[–] [email protected] 112 points 1 month ago* (last edited 1 month ago) (7 children)

I hate to be that person, but what's this new Lemmy trend of posting Twitter screenshots without the link to the original article?

I don't care about some snarky Twitter commenter, I want to see the Octopus make a friend.

[–] [email protected] 1 points 1 month ago* (last edited 1 month ago)

The clinic is, despite its name, not only a stunt but also an audacious legal experiment. The temple claims abortion can be performed as a Satanic sacrament, and therefore is a legally protected religious ritual. They plan to use this to sue to reestablish abortion rights that have been rescinded in other states.

I love this. I love every bit of this, and I never realized the Satanists were "hacking" to this extreme.

Same with the vibes, and even the controversy.

Once a week, Greaves smokes a lot of weed and hosts a virtual movie marathon with viewers tuning in on the temple’s online channel, TST TV. Films can be anything from B horror movies to Christian romcoms — interspersed with scenes Greaves splices together, one minute a topless woman performing in front of bongos, the next dancing meatballs.

“The unbaptism is a way for us to separate from religious trauma from our previous religious lives,” Honey says. To “move more into ourselves and establish our own set of values and establish an opportunity to heal through ritual.”

“I have little to no background or history in Satanism,” he admits.

[–] [email protected] 1 points 1 month ago* (last edited 1 month ago) (1 children)

Well the jist of it is in one sentence:

The net result is that I have transmitted a message into my own past.

Basically, FTL automatically lets you make time machines, and this is bad(TM). It just doesn't make any physical sense, so the consensus is bad things happen (like black holes forming) when you actually push against the speed of light, with very reasonable explanations for why this happens.

The exception is wormholes ,which are theoretically possible "FTL" travel, but only if you are very very careful about where you put them. Otherwise they explode.

[–] [email protected] 1 points 1 month ago* (last edited 1 month ago) (3 children)

See Orion's Arm's explanation:

https://www.orionsarm.com/xcms.php?r=oa-faq&topic=FTL%20in%20OA

And related concept's like a wormhole's failure mode (EG they immediately collapse if ever positioned in a way that allows for actual FTL travel):

https://www.orionsarm.com/eg-article/48545a0f6352a

https://www.orionsarm.com/eg-article/4754be03eb3bc

Orion's Arm is really cool because it's set in the far future, but it tries to limit AI engineering to what's theoretically possible with current physics (just not with current engineering), and they have good explanations for it all. For example, warp drives are a thing, and theoretically plausible, but they do not allow for FTL travel.

 

HP is apparently testing these upcoming APUs in a single, 8-core configuration.

The Geekbench 5 ST score is around 2100, which is crazy... but not what I really care about. Strix Halo will have a 256 -bit memory bus and 40 CUs, which will make it a monster for local LLM inference.

I am praying AMD sells these things in embedded motherboards with a 128GB+ memory config. Especially in an 8-core config, as I'd rather not burn money and TDP on a 16 core version.

view more: next ›