Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Reranking on human pairwise T10 (test)
Loading...
78
EM (%)
Kimi-K2.6
48.88
56.44
64
71.56
Jul 7, 2026
EM (%)
Margin (nats)
Updated 18d ago
Evaluation Results
Method
Method
Links
EM (%)
Margin (nats)
Kimi-K2.6
Inference path=Direct...
2026.07
78
-
SimPO
Inference path=log p(Yes)
2026.07
78
4.28
Base
Inference path=Direct...
2026.07
75.6
0.76
SFT
Inference path=Direct...
2026.07
75.6
-
Base
Inference path=log p(Yes)
2026.07
63.4
0.76
Random
Inference path=Direct...
2026.07
50
-
Random
Inference path=log p(Yes)
2026.07
50
-
Feedback
Search any
task
Search any
task
