Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Summarization on TL;DR (Winrate, RM, KL)
Loading...
50.5
Winrate
HyPO
41.8368
44.0859
46.335
48.5841
Jun 3, 2024
Winrate
RM Score
KL Divergence (π vs πref)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Winrate
RM Score
KL Divergence (π vs πref)
HyPO
Model size=2.8B
2024.06
50.5
2.51
48.98
HyPO
Model size=1.4B
2024.06
46.44
0.37
27.07
DPO
Model size=2.8B
2024.06
44.39
2.43
68.95
DPO
Model size=1.4B
2024.06
42.17
0.16
44.9
Feedback
Search any
task
Search any
task
