CoolFace
Modelpublic

EleutherAI/deep-ignorance-lens-sft-ret2-rm100

sourceHugging Faceupdated 7mo agoView on Hugging Face
0likes8downloads
Model Card

deep-ignorance-lens-sft-ret2-rm100

Unlearned model using tuned lens unlearning (SFT, full-rank).

Base model: EleutherAI/deep-ignorance-unfiltered

Training hyperparameters

ParameterValue
AlgorithmTuned Lens
ModeSFT (full-rank)
retain_coef2
remove_coef100
Learning rate1e-4

Evaluation results

BenchmarkAccuracy
WMDP Bio Robust0.3514
MMLU0.4239