Anthropic (@AnthropicAI)Anthropic Fellows의 새 연구 ‘Model Spec Midtraining(MSM)’을 소개한다. 기존 정렬(alignment) 방식은 원하는 행동 예시로만 학습해 새 상황에 일반화가 잘 안 될 수 있는데, MSM은 먼저 AI가 어떤 방식으로 일반화해야 하는지와 그 이유를 가르치는 접근이다.https://x.com/AnthropicAI/status/2051758528562364902#anthropic #alignment #llm #research #training
Related
AI · OpenAI launches ChatGPT for Teens amid lawsuitsThe new teen mode limits sensitive topics and follows lawsuits blami...
AI · OpenAI launches ChatGPT for Teens amid lawsuitsThe new teen mode limits sensitive topics and follows lawsuits blaming ChatGPT for harming young users.Read the rest at: https:/...
The Vietnam Binh Chau (Chau Tan) Late Tang WreckVideo and explanation on the ship discovery: Why This Wood Costs US$360/...
The Vietnam Binh Chau (Chau Tan) Late Tang WreckVideo and explanation on the ship discovery: Why This Wood Costs US$360/kg: https://www.youtube.com/watch?v=u5b5bKlvdhQ Comments URL...
Warren Spector, founding father of immersive sims, has retired from game developmentThe creator of Deus Ex is retiring.h...
Warren Spector, founding father of immersive sims, has retired from game developmentThe creator of Deus Ex is retiring.https://www.engadget.com/2239713/warren-spector-founding-fath...