Multi-task Embedding Performance on 38 datasets (Total)
49.78Avg ScoreEmbedding Model (EM)
Evaluation results
| Method | Links | |
|---|---|---|
| Embedding Model (EM)Specialization Approach=non-specialized2025.06 | 49.78 | |
| MoTESpecialization Approach=task-specialized expert architecture2025.06 | 3.23 | |
| Instruction-Conditioning (IC)Specialization Approach=instruction-based2025.06 | 2.35 | |
| Specialized Embedding Models (SEM)Specialization Approach=independently trained per task2025.06 | -5.02 |