Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Response-level hallucination detection on RAGTruth (test)
Loading...
74.5
AUC
GASP-threshold (default)
53.076
58.638
64.2
69.762
Jul 5, 2026
AUC
Updated 18d ago
Evaluation Results
Method
Method
Links
AUC
GASP-threshold (default)
Scorer=Qwen2.5-0.5B
2026.07
74.5
GASP-threshold (default)
Scorer=SmolLM2-1.7B
2026.07
74.1
GASP+base (trained)
Scorer=Qwen2.5-0.5B
2026.07
73.1
GASP-trained
Scorer=Qwen2.5-1.5B
2026.07
72.6
GASP+base (trained)
Scorer=SmolLM2-1.7B
2026.07
71.7
GASP-trained
Scorer=SmolLM2-1.7B
2026.07
71.6
GASP-threshold (default)
Scorer=Qwen2.5-1.5B
2026.07
71.3
GASP+base (trained)
Scorer=Qwen2.5-1.5B
2026.07
71.3
GASP-trained
Scorer=Qwen2.5-0.5B
2026.07
70.6
Perplexity
Scorer=Qwen2.5-1.5B
2026.07
62.4
Perplexity
Scorer=Qwen2.5-0.5B
2026.07
58.1
Perplexity
Scorer=SmolLM2-1.7B
2026.07
57.8
Length
Scorer=Qwen2.5-0.5B
2026.07
55.3
Length
Scorer=Qwen2.5-1.5B
2026.07
55.3
Length
Scorer=SmolLM2-1.7B
2026.07
53.9
Feedback
Search any
task
Search any
task