Dissimilar Span Detection on SSD (test)
75F1 Score (Global)LLM
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LLMModel=Claude 3.5 Sonnet, Model Size=Billions, Eval. Time=337.802026.03 | 75 | 89.5 | 64 | |
| Token ClassificationBackbone=roberta-base, Model Size=125M, Eval. Time=1.232026.03 | 69 | 83.8 | 57.4 | |
| EmbeddingBackbone=text-embedding-004, Threshold=0.005, Model Size=Billions, Eval. Time=611.422026.03 | 54.7 | 66.6 | 45.9 | |
| EmbeddingBackbone=all-mpnet-base-v2, Threshold=0.008, Model Size=109M, Eval. Time=11.282026.03 | 46.9 | 52.9 | 42.3 | |
| LIMEBackbone=all-mpnet-base-v2, Threshold=0.001, Model Size=109M, Eval. Time=1981.812026.03 | 46.3 | 78.2 | 22.3 | |
| No-DSD2026.03 | 42.9 | 100 | 0 | |
| SHAPBackbone=all-MiniLM-L6-v2, Threshold=0.010, Model Size=22.7M, Eval. Time=44.472026.03 | 36.6 | 43.4 | 13.1 | |
| Naive-DSD2026.03 | 31.1 | 0.3 | 54.2 |