Sharing new research, models, and datasets from Meta FAIR • Hacker News Board

Re: Sharing new research, models, and datasets from Meta FAIR

#11

Post by echelon » Fri, Dec 13, 2024, 11:23 PM UTC

pkkkzip wrote: post #4

meta has certainly redeemed itself and helping AI become moat-free

Even though Meta doesn't sell I/PaaS, Meta's fitness goes up when AI is in the hands of more players than just Google and OpenAI. Commoditize AI and you create a diverse set of businesses that will reach customers through Meta's platforms.

Re: Sharing new research, models, and datasets from Meta FAIR

#12

Post by puttycat » Fri, Dec 13, 2024, 11:44 PM UTC

Can someone explain how watermarking AI videos voluntarily helps make AI safer?

Re: Sharing new research, models, and datasets from Meta FAIR

#13

Post by causalmodels » Fri, Dec 13, 2024, 11:45 PM UTC

vouaobrasil wrote: post #5

[flagged]

Safety and capabilities research are pretty much two sides of the same coin.

Re: Sharing new research, models, and datasets from Meta FAIR

#14

Post by benatkin » Fri, Dec 13, 2024, 11:46 PM UTC

puttycat wrote: post #12

Can someone explain how watermarking AI videos voluntarily helps make AI safer?

It lets those providing AI video generation services watermark all of their videos. So it isn't intended to by voluntary. You would be left with those services that don't comply with whatever the current Big Tech rules are, like people who used Grok/X.ai to generate images in support of Trump despite Grok/X.ai being inferior. https://arstechnica.com/information-technology/2024/08/musks...

Re: Sharing new research, models, and datasets from Meta FAIR

#15

Post by mupuff1234 » Fri, Dec 13, 2024, 11:52 PM UTC

pkkkzip wrote: post #4

meta has certainly redeemed itself and helping AI become moat-free

It's not redeeming if you still continue with the original sin.

Re: Sharing new research, models, and datasets from Meta FAIR

#16

Post by modeless » Fri, Dec 13, 2024, 11:53 PM UTC

I really hope Dynamic Byte Latent Transformers work out. Death to tokenizers!

Interesting that it's a a hierarchical structure but only two levels of hierarchy. Stacking more levels seems like an obvious direction for further research.

Re: Sharing new research, models, and datasets from Meta FAIR

#17

Post by bbor » Sat, Dec 14, 2024, 12:03 AM UTC

Crazy stuff. Everyone’s covering how exciting all these are (especially LCM and the non-tokenizing-tokenizer), but I have to ask in case anyone’s been paying attention: why are they using the term “advanced machine intelligence”?

My initial thought is that they want to please/distract the doomers, but I’m prolly just self-centered!

Re: Sharing new research, models, and datasets from Meta FAIR

#18

Post by devmor » Sat, Dec 14, 2024, 12:11 AM UTC

bbor wrote: post #17

Crazy stuff. Everyone’s covering how exciting all these are (especially LCM and the non-tokenizing-tokenizer), but I have to ask in case anyone’s been paying attention: why are they using the term “advanced machine intelligence”? My initial thought is that they want to please/distract the doomers, but I’m prolly just self-centered!

I would guess it’s in response to the recent market studies showing that the general public views anything labeled “AI” as a likely scam and untrustworthy.

Re: Sharing new research, models, and datasets from Meta FAIR

#19

Post by intalentive » Sat, Dec 14, 2024, 12:22 AM UTC

Every time I have to clean text I wonder why I haven’t just trained a byte level denoising autoencoder to handle it for me.

Re: Sharing new research, models, and datasets from Meta FAIR

#20

Post by th0ma5 » Sat, Dec 14, 2024, 12:26 AM UTC

vouaobrasil wrote: post #5

[flagged]

It is a shame that this is flagged for being denigrating or negative. The better comment could be to ask where is the documentation for safety? How do we define it? Where are the disclosures about failures, negative results, etc? Perhaps these things are unanswerable, but raising awareness of them is important.