Public KG record
MiniMax Sparse Attention (MSA)
Recorded overview
A fine-grained sparse attention method with a flat block indexer.
An evidence-backed entity from the project knowledge graph; detailed tool metadata is still being added.
Recorded details
- Benchmarks: RULER-13 (MSA: 77.1%)
- Benchmarks: LongBench v2 (MSA: 49.3%)
- Benchmarks: RULER-13（256K）77.1%
- Benchmarks: LongBench v2（thinking）49.3%
- Benchmarks: LongBench v2
- Benchmarks: RULER-13
- Performance: MSA on RULER-13: 77.1%
- Performance: MSA on LongBench v2: 49.3%
Evidence coverage
The KG currently records 7 known information dimensions and has 11 undocumented dimensions, linked to 1 public source articles. These counts describe record coverage, not model performance or verified fit.
Related public records
- Zyphra Hierarchical Attention (ZHA) — competes with
- LongBench v2 — evaluated on
- RULER-13 — evaluated on
- Zyphra Hierarchical Attention (ZHA) — competes with