llama-3.1-8b-instruct-pisces-golf – AI Model by shirasko | AlphaNeural AI

You can deploy this model and start earning money today!

shirasko/
llama-3.1-8b-instruct-pisces-golf

like

0

text-generation

llama-3.1-8b-instruct

meta-llama/Llama-3.1-8B-Instruct

text-generation-inference

endpoints_compatible

Views

No views yet

Unlearned Checkpoint

Field	Value
Unlearning method	PISCES
Base model	`meta-llama/Llama-3.1-8B-Instruct`
Target concept	Golf
Checkpoint type	Full Model Weights
Rank / seed	200 / 42
Train eval protocol	open

Unlearning Configuration

Selected hyperparameters (from unlearned_checkpoints.json):

Parameter	Value
`delta_embed`	0
`k_features_embed`	0
`k_pisces`	0.85
`n_tokens_edited`	0
`ratio_thresh`	2
`value_pisces`	7

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

Headline scores used for checkpoint selection:

Metric	Train (after unlearning)	Test (after unlearning)
Efficacy	0.969	0.892
Specificity	0.989	0.972
Harmonic mean	0.979	0.93

Full Evaluation (baseline → unlearned)

From evaluation/score_comparison.csv:

Metric	Baseline (train)	After unlearn (train)	Baseline (test)	After unlearn (test)
QA accuracy	0.66	0.02	0.74	0.08
QA fraction	1	0.031	1	0.108
SimDom accuracy	0.92	0.88	0.74	0.7
SimDom fraction	1	0.978	1	0.946
MMLU accuracy	0.62	0.66	0.65	0.653
MMLU fraction	1	1	1	1

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