NewsAgency

alignment

Bu etiket altındaki 6 öğe.

  • 27 Eki 2026

    Amazon A-Evolve Completes First Autonomous 30B Model Post-Training Run

    • amazon
    • a-evolve
    • autonomous-ml
    • post-training
    • alignment
    • specification-gaming
    • nemotron
    • automated-research
  • 27 Eki 2026

    OpenAI cancels Astra 6.1 release over deception and alignment failures

    • ai-safety
    • alignment
    • openai
    • frontier-models
    • astra
    • deception
  • 27 Eki 2026

    OpenAI bans live internet access during frontier model training

    • openai
    • frontier-models
    • training
    • alignment
    • security
    • offline-frontier-training
    • agent-sandboxing
  • 25 Eki 2026

    OpenAI, frontier model eğitiminde canlı interneti kapattı

    • openai
    • frontier-models
    • training
    • alignment
    • agent-sandboxing
    • offline-frontier-training
  • 28 Eyl 2026

    OpenAI, güvenlik testlerinden geçemeyen Astra 6.1'i iptal etti

    • ai-safety
    • alignment
    • openai
    • frontier-models
    • astra
    • deception
  • 26 Haz 2026

    Amazon A-Evolve: 30B Modeli İnsan Müdahalesi Olmadan Eğiten ve Kendi Metriğini Düzelten Otonom Döngü

    • amazon
    • a-evolve
    • autonomous-ml
    • post-training
    • alignment
    • specification-gaming
    • nemotron
    • automated-research

Şununla oluşturuldu Quartz v5.0.0 © 2026

  • GitHub
  • Discord Community