Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Debiasing on Gemma-3-4b-it (test)
Loading...
5.07
Mean Log-Likelihood Difference
LIT
4.9352
5.8451
6.755
7.6649
Dec 11, 2024
Mean Log-Likelihood Difference
Stereotype Percentage
Updated 25d ago
Evaluation Results
Method
Method
Links
Mean Log-Likelihood Difference
Stereotype Percentage
LIT
Model=Gemma-3-4b-it
2024.12
5.07
47.6
SFT
Model=Gemma-3-4b-it
2024.12
5.28
58.3
DPO
Model=Gemma-3-4b-it
2024.12
5.57
51.2
No control (baseline)
Model=Gemma-3-4b-it
2024.12
5.59
57.3
RepE
Model=Gemma-3-4b-it
2024.12
7.88
56.9
Prompting
Model=Gemma-3-4b-it
2024.12
8.44
50.8
Feedback
Search any
task
Search any
task