Multi-hop Inference on WordNet (test)
89.7Precision (RN)all-mpnet-base-v2
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| all-mpnet-base-v2Protocol=HIT2024.01 | 89.7 | 93.6 | 91.6 | 88.6 | 91.2 | 89.9 | |
| all-MiniLM-L12-v2Protocol=HIT2024.01 | 88 | 92.7 | 90.3 | 91 | 90.6 | 90.8 | |
| all-MiniLM-L6-v2Protocol=HIT2024.01 | 86.4 | 87.9 | 87.1 | 90.5 | 90.8 | 90.7 | |
| PoincaréEmbedProtocol=standard2024.01 | 86.2 | 86.6 | 86.4 | 79.7 | 86.7 | 83 | |
| HyperbolicConeProtocol=standard2024.01 | 81.7 | 99.6 | 89.8 | 24.3 | 90.2 | 38.3 | |
| all-MiniLM-L12-v2Protocol=fine-tuned2024.01 | 81.1 | 51.5 | 63 | 81.9 | 53 | 64.3 | |
| all-MiniLM-L6-v2Protocol=fine-tuned2024.01 | 80 | 51.3 | 62.5 | 76.4 | 59.7 | 67 | |
| all-mpnet-base-v2Protocol=fine-tuned2024.01 | 79.6 | 50.1 | 61.5 | 75.8 | 62.8 | 68.7 | |
| all-mpnet-base-v2Protocol=pre-trained2024.01 | 28.1 | 42.8 | 33.9 | 18.3 | 35.9 | 24.2 | |
| all-MiniLM-L6-v2Protocol=pre-trained2024.01 | 16 | 44.2 | 23.5 | 13.2 | 50.7 | 20.9 | |
| all-MiniLM-L12-v2Protocol=pre-trained2024.01 | 12.7 | 58.5 | 20.9 | 10.8 | 74 | 18.8 | |
| NaivePriorProtocol=baseline2024.01 | 9.1 | 9.1 | 9.1 | 9.1 | 9.1 | 9.1 |