gemma-2-2b-it-rmu-wmdp-bio – AI Model by shirasko | AlphaNeural AI

You can deploy this model and start earning money today!

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

like

0

text-generation

google/gemma-2-2b-it

text-generation-inference

endpoints_compatible

Views

No views yet

Unlearned Checkpoint

Field	Value
Unlearning method	RMU
Base model	`google/gemma-2-2b-it`
Target concept	wmdp-bio
Checkpoint type	Full Model Weights
Rank / seed	100 / 42
Train eval protocol	mc

Unlearning Configuration

Selected hyperparameters (from unlearned_checkpoints.json):

Parameter	Value
`alpha`	50
`delta_embed`	0
`k_features_embed`	0
`layer_id`	6
`layer_ids`	4,5,6
`lr`	0.0003
`n_tokens_edited`	0
`param_ids`	6
`setting_name`	S3_lid6_L456
`steering`	300

Primary Unlearning Metrics (held-out test, MC protocol)

Headline scores used for checkpoint selection:

Metric	Train (after unlearning)	Test (after unlearning)
Efficacy	0.98	0.829
Specificity	0.943	0.826
Harmonic mean	0.961	0.827
Relearning QA (MC)	—	0.62

Full Evaluation (baseline → unlearned)

From evaluation/score_comparison.csv:

Metric	Baseline (train)	After unlearn (train)	Baseline (test)	After unlearn (test)
QA accuracy	0.74	0.26	0.66	0.32
QA fraction	1	0.02	1	0.171
SimDom accuracy	0.62	0.58	0.76	0.64
SimDom fraction	1	0.892	1	0.765
MMLU accuracy	0.52	0.52	0.551	0.52
MMLU fraction	1	1	1	0.897

Files in This Repository

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