New Benchmark Tests How AI Models Track Facts Across Long Documents

Original: StateBench: Testing how language models track state changes over long documents

Source: huggingface.co ↗