Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Factual Consistency Detection on TofuEval-MeetingBank Span level leakage-clean protocol (test)
Loading...
0.693
AUC
GASP-threshold (default)
0.50788
0.55594
0.604
0.65206
Jul 5, 2026
AUC
Updated 18d ago
Evaluation Results
Method
Method
Links
AUC
GASP-threshold (default)
Model=Qwen-0.5B
2026.07
0.693
GASP-threshold (default)
Model=Qwen-1.5B
2026.07
0.693
GASP-threshold (default)
Model=SmolLM2
2026.07
0.693
GASP+base (trained)
Model=SmolLM2
2026.07
0.68
GASP+base (trained)
Model=Qwen-1.5B
2026.07
0.675
GASP+base (trained)
Model=Qwen-0.5B
2026.07
0.672
GASP-trained
Model=Qwen-0.5B
2026.07
0.656
GASP-trained
Model=Qwen-1.5B
2026.07
0.642
GASP-trained
Model=SmolLM2
2026.07
0.631
Perplexity
Model=Qwen-0.5B
2026.07
0.555
Length
Model=SmolLM2
2026.07
0.535
Perplexity
Model=SmolLM2
2026.07
0.524
Perplexity
Model=Qwen-1.5B
2026.07
0.516
Length
Model=Qwen-0.5B
2026.07
0.515
Length
Model=Qwen-1.5B
2026.07
0.515
Feedback
Search any
task
Search any
task