Insights from Kaggle Grandmasters and Experts on Competitive AI and LLM Frontiers | NVIDIA GTC 2024

Ғылым және технология

Hear from a distinguished panel of Kaggle Grandmasters and experts in computer vision, large language models (LLMs), and data science competitions as they shed light on best practices in AI and the competitive landscape.
Topics include:
▫️ LLM optimization: Learn how to unlock strategies for faster, accurate inference of LLMs and recommender systems, and hear about strategies that won one of the biggest recsys challenges.
▫️ Computer vision: Discover the tools and techniques that top experts are using today.
▫️ Competitive AI: Grain insights into the strategies that lead to top performance in AI challenges.
Panelists:
▫️ Jiwei Liu, Senior Data Scientist, NVIDIA
▫️ Kazuki Onodera, Senior Deep Learning Data Scientist, NVIDIA
▫️ Chris Deotte, Senior Data Scientist, NVIDIA
▫️ Laura Leal-Taixe, Senior Research Manager, NVIDIA
▫️ David Austin, Principal Data Scientist, NVIDIA
Explore more GTC 2024 sessions like this on NVIDIA On-Demand: nvda.ws/3U33qo7
Read and subscribe to the NVIDIA Technical Blog: nvda.ws/3XHae9F
Original GTC 2024 Session: Insights from Kaggle Grandmasters and Experts on Competitive AI and LLM Frontiers [S62792]
#GTC24 #NVIDIA #GTC #AI #GenAI #Generative AI #LLM #LargeLanguageModel #Kaggle #DataScience

Пікірлер: 10

  • @1samson1
    @1samson113 күн бұрын

    The living legends! I will work hard and join you one day!

  • @LeeWoods
    @LeeWoods25 күн бұрын

    It's wild that Bert, Transformers and T5 were developed by a single company - Google. I wonder what other research they have in the pipeline and well they be more conservative about it sharing it.

  • @GNARGNARHEAD
    @GNARGNARHEAD24 күн бұрын

    figures that no one would show up for the best talk of the conference 😆

  • @wobblynl1742
    @wobblynl174224 күн бұрын

    Text is a human way of embedding visual information, makes me wonder are we double encoding or decoding when putting that into an llm embedding space 🤔

  • @qd___8731

    @qd___8731

    24 күн бұрын

    Text also helps explore abstract concept spaces that we can't see, helps to express semantics and details beyond physicality, so I think there'll always be value to text information. Exploring quantum physics as a macroscopic entity, without the ability to intricately describe information would be virtually impossible in any realistic setting.

  • @guanxi99
    @guanxi9919 күн бұрын

    I wonder if there will be ever a LLM that will be able to filter all „you knows“ out of spoken English language. Plus a ML that determines the level of bulshixxx from „you know“ frequency. And the same for „basically“.

  • @wawalkinshaw
    @wawalkinshaw24 күн бұрын

    This made me wiggle.🎉 Thank you!

  • @hussienalsafi1149
    @hussienalsafi114920 күн бұрын

    🥰🥰🥰🥰🥰🥰🥰🥰

  • @senju2024
    @senju202424 күн бұрын

    What is Kaggle? Never heard of it.

  • @thebgEntertainment1

    @thebgEntertainment1

    24 күн бұрын

    very large data platform that hosts competitions for ML/AI/CV. You can find a lot of directly usable notebooks for AI, pretty usefull in my opinion even without doing any competitions.

Келесі