Semantic Retrieval 벤치마크
Semantic Retrieval on Contract Discovery
Soft-F1
- 2019-11-10 — Human baseline: Soft-F1 0.84
| Rank | Model | Soft-F1 | Paper | Code | Year |
|---|---|---|---|---|---|
| 1 | Human baseline | 0.84 | Contract Discovery: Dataset and a Few-Shot Semantic Retrieval Challenge with Competitive Baselines | applicaai/contract-discovery | 2019 |
| 2 | k-NN with sentence n-grams, GPT-2 embeddings, fICA | 0.51 | Contract Discovery: Dataset and a Few-Shot Semantic Retrieval Challenge with Competitive Baselines | applicaai/contract-discovery | 2019 |
| 2 | DBTW, GPT-1 embeddings, fICA | 0.51 | Dynamic Boundary Time Warping for Sub-sequence Matching with Few Examples | 2020 | |
| 4 | LSA baseline | 0.39 | Contract Discovery: Dataset and a Few-Shot Semantic Retrieval Challenge with Competitive Baselines | applicaai/contract-discovery | 2019 |
| 5 | Universal Sentence Encoder | 0.38 | Contract Discovery: Dataset and a Few-Shot Semantic Retrieval Challenge with Competitive Baselines | applicaai/contract-discovery | 2019 |
| 6 | Sentence BERT | 0.31 | Contract Discovery: Dataset and a Few-Shot Semantic Retrieval Challenge with Competitive Baselines | applicaai/contract-discovery | 2019 |