Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Self-correction on WikiText-2 and OpenWebText
Loading...
82.2
NLI Score
Ours-0.6B
69.824
73.037
76.25
79.463
May 14, 2026
NLI Score
RM Score
Overall Score
Levenshtein Distance
Updated 19d ago
Evaluation Results
Method
Method
Links
NLI Score
RM Score
Overall Score
Levenshtein Distance
Ours-0.6B
Parameters=0.6B
2026.05
82.2
1.182
89.9
143.7
Qwen2.5-1.5B
Parameters=1.5B, Type=...
2026.05
82.1
0.355
54.2
1,082.1
Qwen2.5-1.5B-Instruct
Parameters=1.5B, Type=...
2026.05
70.3
1.742
89.7
173.9
Feedback
Search any
task
Search any
task