| --- |
| title: ModelAtlas Community Analyses |
| emoji: 🗺️ |
| colorFrom: blue |
| colorTo: purple |
| sdk: datasets |
| app_file: README.md |
| pinned: false |
| license: cc-by-4.0 |
| tags: |
| - model-analysis |
| - architecture |
| - transformers |
| - community |
| - abliteration |
| --- |
| |
| # ModelAtlas Community Dataset |
|
|
| Community-contributed model architecture analyses from the ModelAtlas project. |
|
|
| ## Dataset Description |
|
|
| This dataset contains detailed architectural analyses of transformer models, contributed by the ModelAtlas research community. Each record provides deep insights into model structure, techniques, and intervention points for research purposes. |
|
|
| ## Access Levels |
|
|
| - **Public**: Basic architectural data |
| - **Contributor**: Intervention mapping (requires contribution history) |
| - **Heretic**: Advanced ablation techniques (requires community approval) |
|
|
| ## Data Schema |
|
|
| ```json |
| { |
| "model_id": "organization/model-name", |
| "schema_version": 3, |
| "config": {"hidden_size": 4096, "num_layers": 36}, |
| "techniques": {"positional_encoding": "rope"}, |
| "provenance": {"organization": "org", "license": "apache-2.0"}, |
| "intervention_map": {"access_controlled": true}, |
| "quality_score": 0.95 |
| } |
| ``` |
|
|
| ## Usage |
|
|
| ```python |
| from datasets import load_dataset |
| |
| # Load public data |
| dataset = load_dataset("RadicalNotionAI/modelatlas-community") |
| |
| # Access level filtering handled by ModelAtlas access control system |
| ``` |
|
|
| ## Contributing |
|
|
| Contribute analyses through the ModelAtlas CLI: |
|
|
| ```bash |
| python atlas.py contribute --setup |
| python atlas.py contribute --submit |
| ``` |
|
|
| ## Community |
|
|
| - **Dashboard**: https://huggingface.co/spaces/RadicalNotionAI/modelatlas-dashboard |
| - **Documentation**: [ModelAtlas GitHub](https://github.com/your-org/ModelAtlas) |
| - **Access Requests**: Through the community dashboard |
|
|
| --- |
|
|
| *Built with ModelAtlas - Architectural Intelligence for AI Research* |
|
|