HindsightBench
Explore score history ↗A black-box audit protocol that measures parametric hindsight leakage in time-indexed LLM decision tasks. It uses date manipulations, memory probes, and six defined metrics, and the paper applies it to 15 models.
- Released
- 2026-07-21
- Institutions
- University College Dublin
- Evaluates
- Language & Knowledge, Robotics & Embodied AI, cs.LG
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- ai-reviewed
HindsightBench paper and code
No reported scores are on record for this benchmark yet.