Who defines "success" for sign language technologies?
Our work with @katwellll, @DanielleKBragg, and @malihealikhani explores this question through a global survey.
More here: https://t.co/kKeg078Sp1
"Measuring How (Not Just Whether) Vision-Language Models Build Common Ground"
We propose a new metric suite to evaluate how VLMs build common ground in dialogue, not just describe images.
🧵👇
✨ Takeaways
VLMs perform well at single-turn tasks but struggle to maintain interactive grounding.
We need training objectives that reward collaborative, incremental dialogue. Not just verbosity or correctness.
🤝 We hope SiLVERScore helps the community build fairer, more robust evaluation pipelines for sign language AI.
📄 Read more: https://t.co/thGL4GNmhA
👩🔬 Authors: @saki__imai, @Merterm, Anthony Sicilia, @malihealikhani
🚀 Excited to share our new paper: “SiLVERScore: Semantically-Aware Embeddings for Sign Language Generation Evaluation”, presented at RANLP 2025.
🧵 A thread on why current metrics fail and how SiLVERScore changes that.
📣 Seeking Deaf Community Insights on Sign Language Technologies
@malihealikhani and I are surveying DHH individuals around the world to amplify global perspectives on sign language technologies.
🌟For each participant, we will donate $5 to the World Federation of the Deaf