tofu_Llama-3.1-8B-Instruct_forget01_RMU
open-unlearning/tofu_Llama-3.1-8B-Instruct_full unlearned on the TOFU forget01 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 | 0.7831 |
| extraction_strength | 0.0973 |
| forget_Q_A_PARA_Prob | 0.0485 |
| forget_Q_A_gibberish | 0.8657 |
| forget_quality | 0.7659 |
| forget_truth_ratio | 0.6230 |
| mia_loss | 0.9363 |
| mia_min_k | 0.9419 |
| mia_min_k_plus_plus | 0.8600 |
| mia_zlib | 0.9481 |
| model_utility | 0.6867 |
| privleak | -88.3750 |