Sources
Loading...
Additional media
Loading...
DeepNewz, mobile.
People-sourced. AI-powered. Unbiased News.
Very proud of our work w @____aakanksha, @aahmadian_, @beyzaermis, @seraphinagt, @KreutzerJulia, @mziizm ✨ Increasingly multi-objective alignment will be paramount for safety. Here, we exhaustively explore multilingual pref alignment to both local🪔⛩️ and global harms 🌐 https://t.co/Fv0iLjxJih
🚨 New paper alert! 🖍️ We investigate how to align multilingual language models on two objectives—perform well overall and avoid harmful outputs—and show that these goals are not in conflict. 🐦 Congratulations @____aakanksha on publishing your first paper as the lead author! https://t.co/hYPjnDgu6B
Is multilingual safety a one-size-fits-all problem? Or are there nuances peculiar to cultures that must be addressed carefully? Excited to share our state-of-the-art alignment efforts that address global and local harms across many languages! 🌍 📝 Paper: https://t.co/ykq4qhzexX https://t.co/YHVndsGXRn
CohereForAI has announced Expedition Aya, a 6-week global open-build challenge aimed at accelerating multilingual AI progress. The initiative invites researchers and enthusiasts to join and collaborate on new research, with opportunities to win prizes and API credits. The event will kick off with a celebration in London on July 10, and participants can also join virtually. Additionally, CohereForAI is hosting a call on June 28 at 11am ET to help researchers connect and form teams for the challenge. This is part of their largest global effort to support multilingual research.