Triplet Evaluation (Ori-Ori) on Intent Semantic Toolkit (test)
51.1TeasyOurs best
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Ours bestbackbone=instructor-large, configuration=-P4, N1,32024.03 | 51.1 | 93.7 | |
| Disable LLMbackbone=instructor-large, LLM-generated data=false2024.03 | 39.3 | 86.2 | |
| Baselinebackbone=instructor-large2024.03 | 23.4 | 87.5 |