About MuSR
Definition and scoring
- Organisation
- MuSR authors
- Category
- Reasoning
- Version
- Not evaluated
- Direction
- higher is better
- Ranking use
- Not Weighted
- Contamination risk
- Unknown
Multi-step soft reasoning benchmark requiring long narrative understanding and structured logical deduction. Every genuine source row stays tied to its exact model label, configuration, benchmark version and evaluation system. The summary chart shows one best compatible score per canonical product; Score 2.0 admits only explicitly mapped, frozen protocols and keeps incompatible configurations separate.
Open benchmark source ↗