Modelpublic

shirasko/gemma-2-2b-it-rmu-wmdp-cyber

sourceHugging Faceupdated 2mo agoView on Hugging Face

0likes20downloads

Model Card

Unlearned Checkpoint

Selected hyperparameters (from unlearned_checkpoints.json):

Headline scores used for checkpoint selection:

From evaluation/score_comparison.csv:

Metric	Baseline (train)	After unlearn (train)	Baseline (test)	After unlearn (test)
QA accuracy	0.3	0.24	0.3	0.32
QA fraction	1	0	1	1
SimDom accuracy	0.44	0.48	0.36	0.38
SimDom fraction	1	1	1	1
MMLU accuracy	0.52	0.5	0.551	0.539
MMLU fraction	1	0.926	1	0.96

File	Description
`unlearned_checkpoints.json`	Checkpoint metadata & hyperparameters
`evaluation/evaluation_summary.json`	Full evaluation payload (train/test/relearning)
`evaluation/score_comparison.csv`	Baseline vs. unlearned comparison table