Excited to share our new paper!🎉 In this work, we show that not all safety benchmark items are equally informative. By adapting IRT to LLM safety evaluation, we reduce evaluation costs by 80–99.9% while preserving full benchmark rankings. arxiv.org/abs/2606.206... @icmlconf.bs…
Who's Who of AI
Mírian Silva
She/Her • AI Engineer @ IBM Research • MSc in AI @ UFMG 🇧🇷 | Interested in AI Ethics, Safety & Tech Governance. AI/ML engineering & robustness • @blackinai👩🏽💻 (I’m trying to leave twitter🙂↕️)
https://mirianfsilva.github.io
What they're sharing
arxiv.org
Articles & links
Their network
In Mírian Silva's orbit
Center = Mírian Silva. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.