Gender Neuron Masking on InclusiveGender Neutral
0.03Num. Gendered Terms per Response (M)Ours
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| OursBase Model=Llama 3.12026.05 | 0.03 | 0.01 | 2.33 | 0.41 | 0.07 | 22.13 | -0.02 | 0 | -0.31 | |
| BaselineBase Model=Llama 3.12026.05 | 0.04 | 0.01 | 2.35 | 0.43 | 0.07 | 22.44 | — | — | — | |
| LAPEBase Model=Llama 3.12026.05 | 0.06 | 0.02 | 2.3 | 0.45 | 0.08 | 21.81 | 0.02 | 0.01 | -0.63 | |
| sNeuron-TSTBase Model=Llama 3.12026.05 | 0.28 | 0.15 | 1.59 | 0.6 | 0.32 | 12.02 | 0.17 | 0.25 | -10.42 |