Jev正式発表:意思決定向けSystem Oneモデル
TypeSafe創業者Diogo Almeidaが、Jev、RLCD訓練手法、ソフトウェア上の意思決定に向けたモデルの位置づけを紹介します。元投稿には2分56秒の発表動画があります。

TypeSafe AIが初のSystem Oneモデル「Jev」を発表。OpenAIのInstructGPTとRLHF研究に参加したDiogo Almeida氏が2年間極秘に開発。
Today, a pretty significant model was released in the AI industry. Its name is Jev. It's the first System One Model from TypeSafe AI, developed in secret for two years by Diogo Almeida, who participated in OpenAI's InstructGPT and RLHF research. The key isn't the model