List your company

Deduplication on RealNews 1M documents

289,229Unique Document CountStandard MinHash
283,025.4284,635.95286,246.5287,857.05Jan 2, 2025

Evaluation results

MethodLinks
Standard MinHashType=Oracle/Brute-force2025.01289,229—
CPU baselineImplementation=SlimPajama, Hardware=64 logical CPU cores2025.01288,68399.8
GPU baselineImplementation=NVIDIA NeMo Curator, Hardware=4 GPUs (V100)2025.01283,72496.6
SEDDHardware=4 GPUs (V100)2025.01283,26496.6