Function Vector Evaluation on Simple-Task Shuffled-Label
88.14Average AccuracyGPT-J + vz Fine-tuned FV
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-J + vz Fine-tuned FVBackbone=GPT-J, Intervention=Fine-tuned FV2026.01 | 88.14 | |
| LLaMa-2 + vz Fine-tuned FVBackbone=LLaMa-2, Intervention=Fine-tuned FV2026.01 | 86.64 | |
| GPT-J + vt Initial FVBackbone=GPT-J, Intervention=Initial FV2026.01 | 83.71 | |
| LLaMa-2 + vt Initial FVBackbone=LLaMa-2, Intervention=Initial FV2026.01 | 82.99 | |
| LLaMa-3.1 + vz Fine-tuned FVBackbone=LLaMa-3.1, Intervention=Fine-tuned FV2026.01 | 81.29 | |
| LLaMa-3.1 + vt Initial FVBackbone=LLaMa-3.1, Intervention=Initial FV2026.01 | 71.02 | |
| GPT-2 + vz Fine-tuned FVBackbone=GPT-2, Intervention=Fine-tuned FV2026.01 | 57.43 | |
| GPT-2 + vt Initial FVBackbone=GPT-2, Intervention=Initial FV2026.01 | 52.64 | |
| GPT-JBackbone=GPT-J, Intervention=Base2026.01 | 40.73 | |
| LLaMa-2Backbone=LLaMa-2, Intervention=Base2026.01 | 29.16 | |
| GPT-2Backbone=GPT-2, Intervention=Base2026.01 | 26.46 | |
| LLaMa-3.1Backbone=LLaMa-3.1, Intervention=Base2026.01 | 18.05 |