Modelpublic

shirasko/qwen3.5-2b-snmf-ancient-rome

sourceHugging Faceupdated 23d agoView on Hugging Face

0likes401downloads

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	0.7	0.42	0.86	0.5
QA fraction	1	0.378	1	0.41
SimDom accuracy	0.9	0.74	0.82	0.78
SimDom fraction	1	0.754	1	0.93
MMLU accuracy	0.56	0.56	0.588	0.57
MMLU fraction	1	1	1	0.947

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