HARE / MEM-SQL: leaderboard
Models answer database questions using previously verified SQL queries. Execution accuracy compares direct generation with distinct memory and schema-linking methods across the published datasets.
Source: arxiv.org.
Benchmark results
Results are published separately for each benchmark or evaluation setting.
- HARE - A-MEM
- HARE - A-MEM - On-Demand Schema Linking
- HARE - BEAVER Origin - A-MEM
- HARE - BEAVER Origin - BIRD-History
- HARE - BEAVER Origin - DAIL-SQL
- HARE - BEAVER Origin - MEM-SQL
- HARE - BIRD Origin - A-MEM
- HARE - BIRD Origin - BIRD-History
- HARE - BIRD Origin - DAIL-SQL
- HARE - BIRD Origin - MEM-SQL
- HARE - BIRD-History
- HARE - BIRD-History - On-Demand Schema Linking
- HARE - DAIL-SQL
- HARE - DAIL-SQL - On-Demand Schema Linking
- HARE - Direct Generation
- HARE - MEM-SQL
- HARE - Spider 2.0 Origin - A-MEM
- HARE - Spider 2.0 Origin - BIRD-History
- HARE - Spider 2.0 Origin - DAIL-SQL
- HARE - Spider 2.0 Origin - MEM-SQL
- HARE - Spider Origin - A-MEM
- HARE - Spider Origin - BIRD-History
- HARE - Spider Origin - DAIL-SQL
- HARE - Spider Origin - MEM-SQL
Interactive version: theaggregate.ai/benchmark?slug=hare-mem-sql-1f4y54i · How It Works · Data refreshed daily, snapshot 2026-11-17.