6 ms·

Facebook AI Research (FAIR) is pretty reputable in the the AI/ML world

Fiahil 5y ago
* with a heavy pinch of salt. I work for a big AI consultancy, but we haven't seen relevant work from Meta for a while. Their non-bayesian stuff is "meh" and often seriously misguided in their business approach. (which makes sense, they're not accustomed to work as consultants on setups with limited data)
- SnooSux 5y ago
  IMO their area of expertise is in software and tooling. The fact that they maintain PyTorch is a big deal and I've used a few of their other libraries like Detectron and torch-points3d and some have become an integral part of my work.
- gumby 5y ago
  So if I understand correctly: they are working in a domain different from yours, thus their work isn’t relevant to you, and even though they are not claiming otherwise their reputation is overrated? Maybe you meant something else, but that’s how your statement reads.
  - ganzuul 5y ago
    Their recent effort to generate training data for spotting stuff that includes unsanctioned narratives comes to mind. https://github.com/facebookresearch/AugLy https://github.com/facebookresearch/AugLy
- oneoff786 5y ago
  I’m a consultant / data scientist for a big management consulting firm. What the hell is a big AI consultancy? As far as I can tell, these do not exist. Why would you expect to see meta doing consulting work at all?
  - whimsicalism 5y ago
    Probably one of those companies trying to sell SMBs a "data platform." ie. like C3.ai ... It seems snake oil adjacent to me, at least at this point.
  - version_five 5y ago
    That caught my attention too. I work on AI consulting and have a reasonable sense of the ecosystem. There are certainly big players that claim "AI" as part of their consulting business, e.g. Deloitte, but they are well removed from any of the research and really just selling dashboards or something with a new name. The GP comments is the kind of thing I would expect a consultant working at a place like that to come out with
- whimsicalism 5y ago
  No clue what a "big AI consultancy" is, but as someone who works in ML on a big feed-driven social networking site, what you are saying is just not true.
- question_away 5y ago
  This isn't even remotely true. wav2vec alone is the most impactful model in speech recognition in a while and their subsequent work in low-resource languages makes it pretty obvious they're working with limited data...
savant_penguin 5y ago
Pretty please don't kill pytorch Thanks
- chalst 5y ago
  What would happen to Pytorch if FB did cut their support? Are the people who depend on it seriously incapable of taking up the load?
  - gumby 5y ago
    > Are the people who depend on it seriously incapable of taking up the load? Probably. First of all most are domain experts not at all systems programmers. That’s the main value of such libraries. And even for those who work close to the iron: plenty depend on, say, the Linux kernel without being able to write patches or even diagnose problems. And few, if any kernel developers have submitted patches to, say, GCC.
    - chalst 5y ago
      There seems to be serious talk about reimplementing the Torchscript core in Julia (i.e., reimplement the C++ part in Julia), something that makes sense to me given the excellent support there already is in Julia for autodiff. Cf. e.g., https://dev-discuss.pytorch.org/t/where-we-are-headed-and-why-it-looks-a-lot-like-julia-but-not-exactly-like-julia/276 https://dev-discuss.pytorch.org/t/where-we-are-headed-and-wh... If these suggestions are credible and don't come from the Facebook team, that might sidestep your concerns.
      - grobclaw 5y ago
        While I would love for that to happen. Quite recently (e.g. [1], also discussed on HN, I believe) the Julia community seemed content with more niche applications, quite deliberately choosing not to challenge the biggies (TensorFlow, Torch, etc.) on deep learning, because the amount of mainainers needed to be competitive was just not there. Is there reason to expect this to change? Maintaining something like Torch and keeping it competitive in terms of speed is a HUGE amount of work for systems development, writing insane numbers of GPU kernels, etc. After having read it, I wouldn't quite yet call the tone of the discussion you linked "serious talk"... [1]: https://discourse.julialang.org/t/state-of-machine-learning-in-julia https://discourse.julialang.org/t/state-of-machine-learning-...
it_does_follow 5y ago
Being reputable and being a vanity division are not mutually exclusive. You could argue that at it's peak Bell labs was a vanity division. That research may have changed the world, but very little of it likely ended up benefiting AT&T in any major way financially. It's telling that once AT&T was broken up Bell labs, while existing in some form for years after, was never reestablished.
- freyr 5y ago
  Bell Labs still exists but is owned by Nokia, who now makes cellular infrastructure. AT&T Labs also exists. But the days of telecom research making rapid and monumental advances peaked decades ago. Nokia or AT&T or Huawei or Ericsson could quadruple their R&D spending and it wouldn’t reestablish the impact of Bell Labs of last century, because it’s simply a much more mature field. Whether it was a vanity division or not, it did serve one practical purpose for AT&T: it presented the company as a benevolent monopoly, spending its profits on developing technologies that benefitted the nation. This helped stave off anti-trust action for a long time.

* with a heavy pinch of salt. I work for a big AI consultancy, but we haven't seen relevant work from Meta for a while. Their non-bayesian stuff is "meh" and often seriously misguided in their business approach. (which makes sense, they're not accustomed to work as consultants on setups with limited data)

SnooSux 5y ago
IMO their area of expertise is in software and tooling. The fact that they maintain PyTorch is a big deal and I've used a few of their other libraries like Detectron and torch-points3d and some have become an integral part of my work.
gumby 5y ago
So if I understand correctly: they are working in a domain different from yours, thus their work isn’t relevant to you, and even though they are not claiming otherwise their reputation is overrated? Maybe you meant something else, but that’s how your statement reads.
- ganzuul 5y ago
  Their recent effort to generate training data for spotting stuff that includes unsanctioned narratives comes to mind. https://github.com/facebookresearch/AugLy https://github.com/facebookresearch/AugLy
oneoff786 5y ago
I’m a consultant / data scientist for a big management consulting firm. What the hell is a big AI consultancy? As far as I can tell, these do not exist. Why would you expect to see meta doing consulting work at all?
- whimsicalism 5y ago
  Probably one of those companies trying to sell SMBs a "data platform." ie. like C3.ai ... It seems snake oil adjacent to me, at least at this point.
- version_five 5y ago
  That caught my attention too. I work on AI consulting and have a reasonable sense of the ecosystem. There are certainly big players that claim "AI" as part of their consulting business, e.g. Deloitte, but they are well removed from any of the research and really just selling dashboards or something with a new name. The GP comments is the kind of thing I would expect a consultant working at a place like that to come out with
whimsicalism 5y ago
No clue what a "big AI consultancy" is, but as someone who works in ML on a big feed-driven social networking site, what you are saying is just not true.
question_away 5y ago
This isn't even remotely true. wav2vec alone is the most impactful model in speech recognition in a while and their subsequent work in low-resource languages makes it pretty obvious they're working with limited data...

savant_penguin 5y ago

Pretty please don't kill pytorch Thanks

chalst 5y ago
What would happen to Pytorch if FB did cut their support? Are the people who depend on it seriously incapable of taking up the load?
- gumby 5y ago
  > Are the people who depend on it seriously incapable of taking up the load? Probably. First of all most are domain experts not at all systems programmers. That’s the main value of such libraries. And even for those who work close to the iron: plenty depend on, say, the Linux kernel without being able to write patches or even diagnose problems. And few, if any kernel developers have submitted patches to, say, GCC.
  - chalst 5y ago
    There seems to be serious talk about reimplementing the Torchscript core in Julia (i.e., reimplement the C++ part in Julia), something that makes sense to me given the excellent support there already is in Julia for autodiff. Cf. e.g., https://dev-discuss.pytorch.org/t/where-we-are-headed-and-why-it-looks-a-lot-like-julia-but-not-exactly-like-julia/276 https://dev-discuss.pytorch.org/t/where-we-are-headed-and-wh... If these suggestions are credible and don't come from the Facebook team, that might sidestep your concerns.
    - grobclaw 5y ago
      While I would love for that to happen. Quite recently (e.g. [1], also discussed on HN, I believe) the Julia community seemed content with more niche applications, quite deliberately choosing not to challenge the biggies (TensorFlow, Torch, etc.) on deep learning, because the amount of mainainers needed to be competitive was just not there. Is there reason to expect this to change? Maintaining something like Torch and keeping it competitive in terms of speed is a HUGE amount of work for systems development, writing insane numbers of GPU kernels, etc. After having read it, I wouldn't quite yet call the tone of the discussion you linked "serious talk"... [1]: https://discourse.julialang.org/t/state-of-machine-learning-in-julia https://discourse.julialang.org/t/state-of-machine-learning-...

it_does_follow 5y ago

Being reputable and being a vanity division are not mutually exclusive. You could argue that at it's peak Bell labs was a vanity division. That research may have changed the world, but very little of it likely ended up benefiting AT&T in any major way financially. It's telling that once AT&T was broken up Bell labs, while existing in some form for years after, was never reestablished.

freyr 5y ago
Bell Labs still exists but is owned by Nokia, who now makes cellular infrastructure. AT&T Labs also exists. But the days of telecom research making rapid and monumental advances peaked decades ago. Nokia or AT&T or Huawei or Ericsson could quadruple their R&D spending and it wouldn’t reestablish the impact of Bell Labs of last century, because it’s simply a much more mature field. Whether it was a vanity division or not, it did serve one practical purpose for AT&T: it presented the company as a benevolent monopoly, spending its profits on developing technologies that benefitted the nation. This helped stave off anti-trust action for a long time.