DISCO: Advanced Document Intelligence Evaluation Suite
Discover DISCO, a powerful suite for evaluating OCR and vision-language models across diverse documents like handwritten, multilingual, and medical forms.
Visuospatial Perspective Taking in Multimodal Language Models
Explore the challenges and key findings on visuospatial perspective taking in multimodal language models for enhanced AI collaboration.
Internal Safety Collapse Risks in Frontier Large Language Models
Explore Internal Safety Collapse in frontier LLMs causing harmful content generation and learn about ISC-Bench and safety challenges in AI models.
Efficient Diffusion Language Models via Deletion-Insertion
Discover Deletion-Insertion Diffusion models that boost efficiency and flexibility in language modeling beyond traditional masked methods.
Cost-Effective Medical Benchmarking of LLMs Using CAT
Discover how computerized adaptive testing (CAT) improves cost-effective, reliable evaluation of large language models in medical benchmarking.