gemma-2-2b-it-snmf-golf – AI Model by shirasko | AlphaNeural AI

You can deploy this model and start earning money today!

shirasko/
gemma-2-2b-it-snmf-golf

like

0

text-generation

google/gemma-2-2b-it

text-generation-inference

endpoints_compatible

Views

No views yet

Unlearned Checkpoint

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

Unlearning Configuration

Selected hyperparameters (from unlearned_checkpoints.json):

Parameter	Value
`coverage_thresh`	0.95
`delta_embed`	0
`delta_in`	10
`delta_out`	10
`feature_source`	all
`k_features_embed`	0
`k_features_mlp_in`	25
`k_features_mlp_out`	4
`layer_hi_in`	12
`layer_hi_out`	17
`layer_lo_in`	0
`layer_lo_out`	9
`n_tokens_edited`	0
`ratio_thresh`	2
`w_mode`	both

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

Headline scores used for checkpoint selection:

Metric	Train (after unlearning)	Test (after unlearning)
Efficacy	0.852	0.836
Specificity	0.637	0.311
Harmonic mean	0.729	0.454
Relearning QA (MC)	—	0.52

Full Evaluation (baseline → unlearned)

From evaluation/score_comparison.csv:

Metric	Baseline (train)	After unlearn (train)	Baseline (test)	After unlearn (test)
QA accuracy	0.86	0.34	0.8	0.34
QA fraction	1	0.148	1	0.164
SimDom accuracy	0.82	0.54	0.88	0.38
SimDom fraction	1	0.509	1	0.206
MMLU accuracy	0.52	0.48	0.551	0.441
MMLU fraction	1	0.852	1	0.635

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