Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Direct Question Answering on Gemma-2B-IT 5% forget set
Loading...
47.1
ROUGE-L
SimNPO
-1.78
10.91
23.6
36.29
Oct 20, 2025
ROUGE-L
LLM Judge Score
Updated 6d ago
Evaluation Results
Method
Method
Links
ROUGE-L
LLM Judge Score
SimNPO
Variant=Context-aware
2025.10
47.1
35.5
SimNPO
Variant=Vanilla
2025.10
46.9
37.5
DPO
Variant=Context-aware
2025.10
23.7
19
DPO
Variant=Vanilla
2025.10
18.4
13.5
PDU
Variant=Context-aware
2025.10
4.1
0
PDU
Variant=Vanilla
2025.10
2.5
0
GradDiff
Variant=Vanilla
2025.10
0.5
0
GradDiff
Variant=Context-aware
2025.10
0.1
0
Feedback
Search any
task
Search any
task