Arabic Labse Matryoshka vs Arabic MiniLM L12 v2 All Nli Triplet
Arabic Labse Matryoshka53.6
Arabic MiniLM L12 v2 All Nli Triplet49.7
| Benchmark | Arabic Labse Matryoshka | Arabic MiniLM L12 v2 All Nli Triplet |
|---|---|---|
| MTEB multilingual v2 · Mean over tasks · multilingual (1037 languages) | 53.6 | 49.7 |
Specs
| Spec | Arabic Labse Matryoshka | Arabic MiniLM L12 v2 All Nli Triplet |
|---|---|---|
| Context | 1K | 1K |
| Weights | Open | Open |
| Released | Jun 2024 | Jun 2024 |
| Dimensions | — | 384 |
| Max input | 0K tokens | 0K tokens |
| Languages | ar | ar |
| Memory (4-bit) | 0.3 GB | 0.1 GB |
| Licence | Apache-2.0 | Apache-2.0 |
| Commercial use | Allowed | Allowed |
What breaks when you switch
Arabic Labse Matryoshka → Arabic MiniLM L12 v2 All Nli Triplet
Re-embed your whole corpus›
Vectors from different models can't be mixed.
Arabic MiniLM L12 v2 All Nli Triplet → Arabic Labse Matryoshka
Re-embed your whole corpus›
Vectors from different models can't be mixed.