Thoughts on llama/meta?
Alibaba team just released an open source LLM that competes with SOTA us labs, deep seek r2 is around the corner. https://x.com/Alibaba_Qwen/status/1916962087676612998
Meanwhile meta's llama models have been hugely criticized as they are literally gaming benchmarks, most benchmarks arent replicated and overall performance was very disappointing
All this while mark zuckerberg just recently went live on podcast and is making bold claims that in 12 months AI will replace top coders with their agentic AI suite
I'm honestly shocked meta is fumbling the bag like this given they have the worlds best talent + insane compute infra. I'm curious to hear others thoughts on meta/llama team
Discussion
(2 comments)- 2
4 days agoSoftware EngineerPeople just expect too much out of LLMs these days. Sure, Meta might have fumbled a little bit on this one, but they basically have infinite money. Just by staying in the game, Meta and Google are effectively just killing the competition. And by the competition, I mean hundreds of startups that could have been potentially building their own LLMs.
I mean, look at any of the smaller foundation model startups. They have insane talent but they just can't play the infinite game that Meta/Google are playing.
- 24 days agoTech Lead/Manager at Meta, Pinterest, Kosei
From a friend at Meta, there's a lot of internal strife and panic about Llama falling behind. (also backed up by blog posts like this)
I do think the public sentiment is that they're behind now, but the benefit of having a singular leader like Mark Zuckerberg is that they can (and will) be patient. Given how much human and capital they can deploy at this, I think they'll find some ways to innovate.
Other Great Discussions
![Senior Software Engineer [E5] at Meta company logo](/_next/image/?url=https%3A%2F%2Ffirebasestorage.googleapis.com%2Fv0%2Fb%2Ftech-career-growth.appspot.com%2Fo%2Fcompany_logos%252Fmeta.jpeg%3Falt%3Dmedia%26token%3Da7a5bcc0-e178-4b5b-b71e-09bf6c6aefae&w=96&q=75){kind=small}
