RAGTruth
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
RAGTruth
92.67Summary Consistency Rate
10
RAGTruth Llama-13B
89.47Recall
10
RAGTruth Llama-7B
92.54Recall
10
RAGTruth 1.0 (test)
0Detection Delay (ARL0=50)
8
RAGTruth QA
95.6AUROC
7
RAGTruth
32.7Hallucination Rate
6
RAGTruth Ex.-F1
83.9Example-F1
5
RagTruth
93.92Score
5
RAGTruth Enhance
100Precision
5
RAGTruth ++
100Precision
5
RAGTruth Summarization Mistral-7b
74.45AUCROC
4
RAGTruth Summarization (Llama-2-13b)
72.9AUCROC
4
RAGTruth Summarization (Llama-2-7b)
73.37AUCROC
4
RAGTruth Span-level conflict vs clean
70.4AUC
3
RAGTruth Span-level baseless vs clean
64.1AUC
3
RAGTruth (test)
57.4F1 Score
3
RAGTruth++
91.2Precision
2
RAGTruth-Enhance
72.7Precision
2