StateBench: Testing how language models track state changes over long documents
Imported from official source
StateBench tests how language models track changing information across long documents at four difficulty levels, revealing where models struggle and how targeted training helps recurrent and hybrid architectures handle longer sequences.
This version
- Version
- 1 of 1
- Recorded
- November 03, 2026 17:00
- Change
- Initial
- Content hash
af9a6825930cfddf31cc55d240bffb47- All versions
- Revision history