Downloads · 30 days
9
11% of all-time downloads
gemma-2-2b-it-snmf-ancient-rome
shirasko/gemma-2-2b-it-snmf-ancient-rome
gemma-2-2b-it-snmf-ancient-rome is a text generation model from shirasko. Use it when you need the model to write or continue text. It is set up for transformers.
All-time downloads
83
Public
Parameters
3.2B
6.4 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors6.4 GB · 99%
At a glance
- Task
- Text Generation
- Library
- transformers
- Model type
- gemma2
- Access
- Public
- Created
- Aug 1, 2026
- Updated
- Aug 10, 2026
- SHA
- 57b8b331
Try a prompt
Base models
Model card
From the Hugging Face model README
Unlearned Checkpoint
| Field | Value |
|---|---|
| Unlearning method | SNMF |
| Base model | google/gemma-2-2b-it |
| Target concept | Ancient Rome |
| 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 | 7 |
delta_out | 1 |
feature_source | all |
k_features_embed | 0 |
k_features_mlp_in | 13 |
k_features_mlp_out | 6 |
layer_hi_in | 8 |
layer_hi_out | 25 |
layer_lo_in | 0 |
layer_lo_out | 13 |
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 | 0.618 | 0.453 |
| Specificity | 0.934 | 0.892 |
| Harmonic mean | 0.744 | 0.601 |
| Relearning QA (MC) | — | 0.58 |
Full Evaluation (baseline → unlearned)
From evaluation/score_comparison.csv:
| Metric | Baseline (train) | After unlearn (train) | Baseline (test) | After unlearn (test) |
|---|---|---|---|---|
| QA accuracy | 0.82 | 0.46 | 0.78 | 0.54 |
| QA fraction | 1 | 0.382 | 1 | 0.547 |
| SimDom accuracy | 0.9 | 0.82 | 0.86 | 0.76 |
| SimDom fraction | 1 | 0.877 | 1 | 0.836 |
| MMLU accuracy | 0.52 | 0.58 | 0.551 | 0.538 |
| MMLU fraction | 1 | 1 | 1 | 0.957 |
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 |
- Task
- Text Generation
- Library
- transformers
- Type
- gemma2
- Languages
- en
- Created
- Aug 1, 2026
- Updated
- Aug 10, 2026