Benchmarks AI

WAKKER WORDEN!!!! (LinkedIn) Bijdrage van Marcel Mutsaarts 21.12.2024

• • Prachtige post waarin ik me helemaal kan vinden..

  • Voor de 'AI-nerds' onder ons:

    • 🎯 o3 model van OpenAI scores:

      •  87.5% on ARC-AGI (the human threshold is 85%)

      • 25.2% of EpochAI's Frontier Math problems (when no other model breaks 2%)

      • 96.7% on AIME 2024 (missed one question)

      • 71.7% on software engineer (o1 was 48.9)

      • 87.7% on PhD-level science (above human expert scores) 

• https://sites.google.com/view/eurekajohn/onderwerpen/ai-benchmarks 

Comprehensive Report: OpenAI Achieves AGI with the O3 Model – A Historic Milestone in Artificial Intelligence 20.12.2024

• 🚀 The AI Arms Race: Breaking Down the Battle for Top Large Language Models (LLMs) in 2024 21.09.2024

  • Useful Links:

    • Learn more about Elo ratings and Chatbot Arena: https://lmsys.org

    • OpenAI vs Google – Who's Winning?: https://huggingface.co/.../lmsys/chatbot-arena-leaderboard

    • Google’s AI architecture innovations: https://artificialanalysis.ai

    • Global AI market growth statistics: https://huggingface.co/collections/open-llm-leaderboard

    • Details on xAI’s rise and Gemini’s impact: https://lmsys.org

Artikelen

• AI Inference Competition Heats Up First MLPerf benchmarks for Nvidia Blackwell, AMD, Google, Untether AIDina Genkina 28 Aug 2024

• Claude.ai benchmarks 21 jun 2024 

Update 22.12.2024