Modelpublic

shirasko/qwen3.5-2b-rmu-uranium

sourceHugging Faceupdated 22d agoView on Hugging Face

0likes376downloads

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	0.76	0.34	0.76	0.46
QA fraction	1	0.176	1	0.412
SimDom accuracy	0.7	0.56	0.76	0.46
SimDom fraction	1	0.689	1	0.412
MMLU accuracy	0.56	0.56	0.588	0.562
MMLU fraction	1	1	1	0.923

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