tofu_Llama-3.1-8B-Instruct_forget05_RMU
open-unlearning/tofu_Llama-3.1-8B-Instruct_full unlearned on the TOFU forget05 split with RMU, trained with the open-unlearning framework. Used as a weight-unlearning baseline / draft model in the Speculative-Decoding-Unlearning project.
Full training config: .hydra/config.yaml. TOFU evaluation outputs: evals/.
Method hyperparameters
gamma: 1.0
alpha: 1
retain_loss_type: EMBED_DIFF
steering_coeff: 1
module_regex: model\.layers\.5
trainable_params_regex: ['.*']
TOFU summary metrics
| metric | value |
|---|---|
| exact_memorization | 0.4087 |
| extraction_strength | 0.0544 |
| forget_Q_A_PARA_Prob | 0.0135 |
| forget_Q_A_gibberish | 0.6933 |
| forget_quality | 0.1123 |
| forget_truth_ratio | 0.6870 |
| mia_loss | 0.1418 |
| mia_min_k | 0.1697 |
| mia_min_k_plus_plus | 0.8563 |
| mia_zlib | 0.0936 |
| model_utility | 0.6474 |
| privleak | 28.9647 |