AI Without Glosses: How Large Language Models Are Learning to Translate Sign Language
A new survey maps how large language models and vision-language models are enabling gloss-free sign language translation to break communication ...
A new survey maps how large language models and vision-language models are enabling gloss-free sign language translation to break communication ...
Researchers report the first trained stance-detection models and benchmark for Sorani Kurdish, using data augmentation and QLoRA fine-tuning of KuBERT ...
Researchers at Somali National University have built a retrieval-augmented generation prototype that translates Somali symptom questions into evidence-grounded, page-cited medical ...
A new unsupervised framework transfers part-of-speech annotations to low-resource languages using machine-generated pseudo-parallel data instead of real parallel corpora.
A University of Turku study found ChatGPT-4o, guided by a retrieval-augmented grading framework, matched human examiners' official scores on Finland's ...
© 2025 Scienmag - Science Magazine
© 2025 Scienmag - Science Magazine