NVIDIA's Nemotron-4's is totally insane for synthetic data generation

Поділитися
Вставка
  • Опубліковано 29 вер 2024
  • the nvidia nemotron-4 340 billion parameter models are brand new open source models from nvidia that you can use to generate your own synthetic data. in this video chris shows you how to get started with nvidia cloud and nvidia nim services. using nvidia cloud you can use all the major open source and commericial models such as llama3, mistral, google and ibm granite models. in this video we focus on the new nvidia nemotron instruct and reward models showing how you can use the models to generate synthetic data, and judge the outputs using the reward model. even if you're not interested in training models, the nvidia cloud is very fast for training and inference of ai models.

КОМЕНТАРІ • 17

  • @pratiknarendraraut6889
    @pratiknarendraraut6889 3 місяці тому +4

    igrore the dude below, awesome content , keep going .

  • @sidharthrao9005
    @sidharthrao9005 3 місяці тому +2

    Is there any possiblities to fine-tune, quantise or distill the model into a smaller model to be run locally? I urgently need this for a project I'm doing.

    • @chrishayuk
      @chrishayuk  2 місяці тому

      i'm not aware of anyone that has done this tbh, it's quite a large model. and it'd be a challenge to get it small enough

  • @erfanshayegani3693
    @erfanshayegani3693 2 місяці тому +1

    Your videos help me a lot stay up-to-date with the latest stuff. This video for example, led me to these new reward models and datasets that have been released only a few days ago!
    I think if you can also have some 10 min videos weekly to summarize the important releases and new stuff it would be great!
    Many thanks for the great content again!

    • @chrishayuk
      @chrishayuk  2 місяці тому

      Sadly, 10 mins weekly summary videos isn’t my style. Check out the IBM mixture of experts podcast though, I’m usually on every 4 weeks or so

  • @bamh1re318
    @bamh1re318 2 місяці тому +1

    what kind of hardware requirement do I need to run this? Thanks!

    • @chrishayuk
      @chrishayuk  2 місяці тому +1

      more than most can afford, it really needs to be run on cloud. i think it's like a cluster of 8 x H100's or something to run it, not sure off the top of my head

  • @richsoftwareguy
    @richsoftwareguy 3 місяці тому

    Maybe you need new glasses. Your eyes keep getting stuck together, hella annoying.

    • @everyhandletaken
      @everyhandletaken 3 місяці тому +3

      How respectful of you. There are other channels, if you can't cope.

    • @chrishayuk
      @chrishayuk  3 місяці тому

      lol, tell me about it, I have a light sensitivity hence why I wear the glasses. I don’t think anything but my sunglasses will fix that, sadly

    • @WillJulian-Vicary
      @WillJulian-Vicary 3 місяці тому +3

      @@chrishayuk Ignore the troll, thought your video was great - thanks for sharing.

    • @chrishayuk
      @chrishayuk  3 місяці тому

      thank you, appreciate it

    • @chrishayuk
      @chrishayuk  3 місяці тому

      thank you, appreciate it