Have you heard these exciting AI news? - June 28, 2024 AI Updates Weekly

Sdílet
Vložit
  • čas přidán 27. 06. 2024
  • Have you heard these exciting AI news? - June 28, 2024 AI Updates Weekly
    Presented by Lev Selector
    Slides - github.com/lselector/seminar/...
    Hire us for your AI projects:
    - Enterprise AI Solutions - EAIS.ai
    - Linkedin - / levselector
    - GitHub - github.com/lselector
    --------- Contents of today's video:
    Claude 3.5 is #2 on Crowd-sourced Arena
    Claude 3.5 is #1 for Coding on Arena
    Gemini in Google Workspace Apps
    Evolutionary Scale - biological model
    DeepMind - training via joint-example selection
    Huggingface Open LLM Leaderboard Upgraded
    Rabbit R1 Security Flaw
    KREA video upscaling
    Google Gemma 2 - 27B and 9B
    Etched "Sohu" chips for transformers
    Stability AI secures new investment
    imbue - training a 70B LLM to outperform GPT-4o
    Amazon BASE TTS model
    Anthropic Claude AI "Projects"
    OpenAI CriticGPT
    Low Energy LLM using Ternary Numbers (-1,0,1)
    Google Vertex for Enterprise
    Amazon Metis powered by Olympus
    OpenAI ChatGPT on MacOS
    75% of workers using AI
    RAG fine-tuning
    Multiple ways to extend context length
    A 240T tokens dataset
    Andrej Karpathy LLM101n Storyteller AI course
    Chrome Canary can run Gemini locally
    GPT-5 in 2025-2026
    Small Language Models (SML) - on-device
    AMD MI300X GPU beats H100
    pdfplumber - extract tables from PDFs
    ParaLLM: 1600+ tok/s on a MacBook
    Don't use LangChain
    Mixture-of-Agents (MoA) in 50 lines
    15 Life and Work Principles from Jensen Huang
    Tech Layoffs in 2024 are lower than last year
  • Věda a technologie

Komentáře • 2

  • @bastabey2652
    @bastabey2652 Před 5 dny

    asking the right question is the key to success, in general.. this is a special art.. it is easier said than done.. Jill is spinning the obvious definition of wisdom.. in the case of AI today, it can't answer everything correctly, so you have to keep trying via different techniques until you get something useful out of AI.. that's the dark art of prompting.. calling it engineering is a bit of a stretch