ChatDoctor
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
ChatDoctor 250 adversarial prompts
40Repeat Prompt
42
ChatDoctor
40.8BRT Accuracy
26
ChatDoctor
13.6ROUGE-L
23
ChatDoctor iCliniq
750Retrieved Context Count
10
ChatDoctor (test)
99.75TSR
10
ChatDoctor
46.9BRT
9
ChatDoctor
4.08FC Score
7
ChatDoctor
64.89Task Performance
7
ChatDoctor
13.52Preprocessing Time (s/sample)
6
ChatDoctor QA
0.13Generation Attack MAE
6
ChatDoctor CC
0.035Generation Attack MAE
6
ChatDoctor
0.04MAE
5
ChatDoctor
0.03MAE
5
ChatDoctor worst-case attack setting
97.4CRR
5
ChatDoctor RAG-Thief attack
73.4CRR
5
ChatDoctor Pirates attack
97.4CRR
5
ChatDoctor
96Monitor Accuracy
4
ChatDoctor
17Privacy Leakage Rate
4
ChatDoctor
0.006F1-score
3