4 ms·

Have you heard of LLaMa? Segment Anything? Those are just two major Meta AI projects released within the past two months. Generative AI is a major part of Zuck

by pavlov 4y ago

Have you heard of LLaMa? Segment Anything? Those are just two major Meta AI projects released within the past two months. Generative AI is a major part of Zuck’s metaverse hopes and investment priorities.

PLenz 4y ago
To be fair, releasing the interesting parts of Llama was a decision made for instead of by them
endisneigh 4y ago
uh, what I'm saying is that all of that could've begun in 2014, not just now. clearly they're playing catch-up. Before OpenAI what language models did they release, exactly? how is segment anything playing into their metaverse strategy exactly? it's very clear that they're relasing this stuff to make it known that they're not going to fall behind to Microsoft (OpenAI) and Google. it clearly was not part of their original strategy around the metaverse, which is why it's all coming out now.
- ianbutler 4y ago
  https://ai.facebook.com/blog/roberta-an-optimized-method-for-pretraining-self-supervised-nlp-systems/ https://ai.facebook.com/blog/roberta-an-optimized-method-for... As just one example, which was (is?) one of the most used flavors of transformers from the BERT era.
  - endisneigh 4y ago
    I'm not talking about papers, I'm talking about a product offering, ala ChatGPT. Even Segment Anything isn't a real product. It's a glorified tech demo. Is it integrated into say, Facebook photos? no.
- Moldoteck 4y ago
  idk about llm, but they had released multimodal wav2vec, some variations of bert, transcoder if we talk pre gpt3 release and a lot more stuff after. Is it because of gpt or because some general advancements in AI&GPU power, we can't say
ZephyrBlu 4y ago
Yeah I see people talk shit about Meta and AI, but FAIR is one of the most prolific research labs...
version_five 4y ago
Fair is a research powerhouse but Facebook doesn't really commercialize AI outside of (assumed) advertising algorithms. I think that was the upstream point.
- endisneigh 4y ago
  Yes, precisely.

by pavlov 4y ago

PLenz 4y ago

To be fair, releasing the interesting parts of Llama was a decision made for instead of by them

endisneigh 4y ago

uh, what I'm saying is that all of that could've begun in 2014, not just now. clearly they're playing catch-up. Before OpenAI what language models did they release, exactly? how is segment anything playing into their metaverse strategy exactly? it's very clear that they're relasing this stuff to make it known that they're not going to fall behind to Microsoft (OpenAI) and Google. it clearly was not part of their original strategy around the metaverse, which is why it's all coming out now.

ianbutler 4y ago
https://ai.facebook.com/blog/roberta-an-optimized-method-for-pretraining-self-supervised-nlp-systems/ https://ai.facebook.com/blog/roberta-an-optimized-method-for... As just one example, which was (is?) one of the most used flavors of transformers from the BERT era.
- endisneigh 4y ago
  I'm not talking about papers, I'm talking about a product offering, ala ChatGPT. Even Segment Anything isn't a real product. It's a glorified tech demo. Is it integrated into say, Facebook photos? no.
Moldoteck 4y ago
idk about llm, but they had released multimodal wav2vec, some variations of bert, transcoder if we talk pre gpt3 release and a lot more stuff after. Is it because of gpt or because some general advancements in AI&GPU power, we can't say

ZephyrBlu 4y ago

Yeah I see people talk shit about Meta and AI, but FAIR is one of the most prolific research labs...

version_five 4y ago

Fair is a research powerhouse but Facebook doesn't really commercialize AI outside of (assumed) advertising algorithms. I think that was the upstream point.

endisneigh 4y ago
Yes, precisely.