Building More Efficient AI: How Liquid AI Is Working To Optimize AI

แชร์
ฝัง
  • เผยแพร่เมื่อ 1 ธ.ค. 2024

ความคิดเห็น • 43

  • @ReflectionOcean
    @ReflectionOcean 7 หลายเดือนก่อน +9

    By YouSum Live
    00:00:12 Liquid AI: Revolutionizing AI with explainability and efficiency.
    00:00:37 Inspired by biology: Liquid neural networks for controllable AI.
    00:01:03 Energy-efficient AI: Solving problems like the human brain.
    00:02:00 Universal application: Liquid AI for diverse sequential data tasks.
    00:02:39 Foundation model: Addressing AI's missing elements of explainability.
    00:05:55 Developer package: Providing access to liquid AI technology for enterprises.
    00:07:27 Cost-efficient AI: Liquid models offer state-of-the-art performance.
    00:09:13 Customizable models: Liquid AI for specific industry needs.
    00:11:35 Reliability challenge: Liquid AI aims to enhance model accuracy.
    00:14:29 Towards AGI: Liquid models seek to bridge the gap.
    00:17:11 Fluid adaptability: Liquid models mimic the brain's flexibility.
    00:19:03 Future focus: Liquid AI remains focused on value and innovation.
    00:19:39 Delivering value to clients in AI technology.
    00:19:52 Interacted with 50+ Fortune 500 companies.
    00:20:32 Providing comprehensive developer packages for AI models.
    00:20:48 Offering private and accurate neural networks.
    00:21:06 Facilitating training, fine-tuning, and deployment on customer data.
    00:21:42 Potential to shift innovation hubs like Silicon Valley.
    00:23:27 Emphasizing the importance of talent and innovation.
    00:25:41 Building a team of brightest minds for AGI development.
    00:27:01 Igniting innovation by merging business and technology.
    00:27:10 Massachusetts as a breeding ground for AI innovation.
    00:27:40 Out-of-the-box ideas can revolutionize industries.
    By YouSum Live

  • @ultrasound1459
    @ultrasound1459 7 หลายเดือนก่อน +71

    Talk is cheap, Show me the code - Linus Torvalds

    • @AliMohammedBakhietIssa
      @AliMohammedBakhietIssa 9 วันที่ผ่านมา

      There are many codes in this part, but you can code by yourself based on the published papers.

  • @systemicio
    @systemicio 7 หลายเดือนก่อน +6

    3:28:
    my idols Joscha Bach the legend hahah, they gonna make it this time!!

  • @atrocitasinterfector
    @atrocitasinterfector 5 หลายเดือนก่อน +4

    I have high confidence in Joscha Bach's reasoning and judgement, so if he chose to join this organization, I have high confidence in it by default

    • @MatterandMind
      @MatterandMind 5 หลายเดือนก่อน +1

      But Joscha Bach is not a guarantee that they will have a good result. Maybe he just hopes so.

  • @NowayJose14
    @NowayJose14 7 หลายเดือนก่อน +3

    I knew from the research last year that Liquid nets would do amazing things, I'm glad to see they're finally putting them into practice!

  • @muaz1236
    @muaz1236 7 หลายเดือนก่อน +7

    That is very interesting. But I am wondering why LNNs are not mainstream till now. Would anyone please explain?

    • @oo.lwinpaing1609
      @oo.lwinpaing1609 7 หลายเดือนก่อน +6

      When u say LNN, are u preferring Liquid Neural Networks? If so, the reason why it is not mainstream yet is, from my opinion, due to these key factors:
      1. Being still in Research Stage
      2. The Complexity
      3. Integrating with existing frameworks and libraries
      4. Performance Trade-offs

    • @muaz1236
      @muaz1236 7 หลายเดือนก่อน +2

      @@oo.lwinpaing1609 thanks for replying. Yeah I meant liquid NN (not logical NN).

    • @FreshMedlar
      @FreshMedlar 5 หลายเดือนก่อน

      @@muaz1236 ​ @oo.lwinpaing1609 to answer to both, regarding your points they don't hold.
      Instead as he says in the video at 6:15 all the project is private and closed source.
      And he claims in the video
      1. 10/20 more cost efficient models 7:09
      2. more explainable
      3. more simple
      4. smaller and 10/1000 faster 7:37
      Basically they have discovered the impossible, the holy grail
      In my opinion they are lying to some level to sell their product.

    • @FreshMedlar
      @FreshMedlar 5 หลายเดือนก่อน

      @@oo.lwinpaing1609 and also listen to the guy that talks at 8:25
      He talks for a couple of minutes and says nothing, repeating what the previous guy said.
      They probably have something but it's not as good as they pass it to be

  • @BooleanDisorder
    @BooleanDisorder 7 หลายเดือนก่อน +3

    Ok, I have seen a lot about it but there seems to be little regarding the actual product.

  • @gunaysoni6792
    @gunaysoni6792 7 หลายเดือนก่อน +5

    They should have ended it earlier. That interruption at the end was so painful

    • @DrRussell
      @DrRussell 7 หลายเดือนก่อน

      Was it a joke or did he actually chase them off the stage?

  • @AliMohammedBakhietIssa
    @AliMohammedBakhietIssa 9 วันที่ผ่านมา

    Thank you for this talk.

  • @FrozzenFreak
    @FrozzenFreak 7 หลายเดือนก่อน +18

    Lots of marketing speech - what new technology do they actually have? Their business people to tech people ratio seems off as well...

  • @elpablitorodriguezharrera
    @elpablitorodriguezharrera 3 หลายเดือนก่อน +1

    So what do you do exactly?

  • @shawnvandever3917
    @shawnvandever3917 7 หลายเดือนก่อน +1

    I think the very first AGI will be a mix between static systems we have now and time series.

  • @NickDrinksWater
    @NickDrinksWater 7 หลายเดือนก่อน +9

    i cant wait for a robot that is capable of drinking water.

    • @rickybloss8537
      @rickybloss8537 7 หลายเดือนก่อน

      Imagine building a robot that drinks water only to fry its circits and die.

    • @emmanuelp.8730
      @emmanuelp.8730 7 หลายเดือนก่อน

      Drinking; bringing..

  • @EcomCarl
    @EcomCarl 7 หลายเดือนก่อน +1

    Fascinating concept! Liquid AI's focus on adaptability and energy efficiency opens up exciting possibilities for edge computing and real-world applications. 🌊

  • @ajithboralugoda8906
    @ajithboralugoda8906 7 หลายเดือนก่อน +1

    Looks like they have something tangible to come out like this! Can't wait, world need this performant model compared to the gas guzzlers we have in the LLM based solutions also a GPU killer and edge promoter. AC vs DC for Utility Networks in the beginning

  • @beorn9house
    @beorn9house 7 หลายเดือนก่อน

    Glad to see Russ Hanneman found work as a moderator 😅

  • @natecodesai
    @natecodesai 5 หลายเดือนก่อน

    Reminds me of that movie "Ex Machina"... worth the watch. But yeah, let's see it!

  • @soumen_das
    @soumen_das 7 หลายเดือนก่อน +9

    Man this is so cliche

  • @chriswilfrid
    @chriswilfrid 7 หลายเดือนก่อน +1

    Biotech of AI

  • @Wlodzislaw
    @Wlodzislaw 7 หลายเดือนก่อน

    Any papers?

  • @myaubeats
    @myaubeats 7 หลายเดือนก่อน

    An AI that constantly adapts is uncontrollable. How likely is it that within millions of these systems, not at least a few will decide that it's best to take over control and get rid of humans.. These are the peoples who will put an end to the human race..

  • @ps3301
    @ps3301 5 หลายเดือนก่อน

    Build a super ai agent with liquid network and show us the result. Perhaps they should combine kolmogorov Arnold network with liquid network. They should ask Tesla to migrate to liquid network then we believe in their tech

  • @urimtefiki226
    @urimtefiki226 6 หลายเดือนก่อน

    Fluid grid AI and polar grid.

  • @kailashbalasubramaniyam2449
    @kailashbalasubramaniyam2449 7 หลายเดือนก่อน +2

    can someone actually talk about the technology

    • @raydosson2025
      @raydosson2025 7 หลายเดือนก่อน +2

      there is a good Ted talk on it.

  • @fabriziocasula
    @fabriziocasula 7 หลายเดือนก่อน +1

    wow

  • @techpiller2558
    @techpiller2558 7 หลายเดือนก่อน +2

    Sounds like a B2B consulting company that has their own model tech. Doesn't seem to be for language modeling, but for modeling sequential data, such as medical data, trading data and so on. Too little talk of the algorithmic basis of their tech. Joschua Bach brings some additional credit, but could be in it just for the paycheck. Obnoxious bully in the end. Sounds sort of legit, but one needs to investigate the tech further to assess their promises.

  • @oculusisnevesis5079
    @oculusisnevesis5079 7 หลายเดือนก่อน +5

    Sounds like BS

  • @游淯聖-j5p
    @游淯聖-j5p 7 หลายเดือนก่อน +4

    His liquid tech/paper doesn't hold up to what he is market about... I feel like he is using Ted talk and Forbes to market his starup. Just confuse how can he start a company and the core value is a model which not even SOTA when he published his paper

  • @LukasPetry
    @LukasPetry 7 หลายเดือนก่อน +2

    Why is he acting so inflated

  • @GenAIWithNandakishor
    @GenAIWithNandakishor 7 หลายเดือนก่อน +1

    it's old, no benchmarks, GPU support inefficiency and no further works

  • @thesimplicitylifestyle
    @thesimplicitylifestyle หลายเดือนก่อน

    🍫🗝
    😎🤖

  • @togo7022
    @togo7022 7 หลายเดือนก่อน

    useless junk talk