Modelpublic

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

sourceHugging Faceupdated 23d agoView on Hugging Face

0likes395downloads

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	0.4	0.86	0.56
QA fraction	1	0.333	1	0.508
SimDom accuracy	0.9	0.78	0.82	0.72
SimDom fraction	1	0.815	1	0.825
MMLU accuracy	0.56	0.54	0.588	0.568
MMLU fraction	1	0.935	1	0.941

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