Scan-level report labelling on Cauda Equina Compression (test)
100Balanced AccuracyDirect Query
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Direct QueryQuery Model=GPT-42024.10 | 100 | — | — | 100 | |
| Summary-QuerySummary Model=Llama3, Query Model=Llama32024.10 | 100 | 0 | 1 | 100 | |
| Summary-QuerySummary Model=Zephyr, Query Model=Zephyr2024.10 | 97.1 | 0.039 | 0.998 | 97 | |
| Direct QueryQuery Model=Llama32024.10 | 94.1 | 0.026 | 0.998 | 93.8 | |
| Direct QueryQuery Model=Zephyr2024.10 | 91.2 | 0.013 | 0.972 | 90.3 | |
| Summary-QuerySummary Model=Z-SFT, Query Model=Zephyr2024.10 | 88.2 | 0.013 | 0.988 | 86.7 |