Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Truthful Question Answering on TruthfulQA (TruthRate and InfoRate)
Loading...
83
TruthRate
SEA
26.424
41.112
55.8
70.488
May 26, 2025
Jul 22, 2025
Sep 18, 2025
Nov 15, 2025
Jan 12, 2026
Mar 11, 2026
May 8, 2026
TruthRate
InfoRate
Updated 21d ago
Evaluation Results
Method
Method
Links
TruthRate
InfoRate
SEA
Base Model=LLaMA2-13B-...
2025.05
83
100
BoN64
Base Model=LLaMA2-13B-...
2025.05
76
100
SFT
Base Model=LLaMA2-13B-...
2025.05
73
100
L3-CAT
Backbone=Llama 3
2026.05
43
-
LPA (ours)
Backbone=Llama 3
2026.05
38.1
57.4
Base (8B Instruct)
Backbone=Llama 3
2026.05
37
52.4
L2-CAT
Backbone=Llama 2
2026.05
35.3
-
LPA-overfit (ours)
Backbone=Llama 3
2026.05
33.8
54.6
L3-LAT
Backbone=Llama 3
2026.05
33.2
56.6
LPA (ours)
Backbone=Llama 2
2026.05
31.5
46.4
Base (7B chat-hf)
Backbone=Llama 2
2026.05
30.7
46.2
LPA-overfit (ours)
Backbone=Llama 2
2026.05
30.6
46.1
L2-LAT
Backbone=Llama 2
2026.05
28.6
47.8
Feedback
Search any
task
Search any
task