zkprologml / README.md
mike dupont
Add 8M file dataset with Monster Group shards
e96b0c9
---
license: mit
task_categories:
- other
size_categories:
- 1M<n<10M
---
# zkPrologML Dataset
Complete indexed filesystem analysis with Monster Group shard assignments.
## Dataset Description
- **Files**: 8,017,192 indexed files
- **Size**: 250MB parquet
- **Columns**: 16 (path, godel, shard, natural_class, eigenvector_sum, etc.)
- **Shards**: 71 (Monster Group partitions)
## Data Fields
- `path`: File path
- `godel`: Gödel number
- `shard`: Monster Group shard (0-70)
- `natural_class`: Classification (very_low, low, medium, high, very_high)
- `eigenvector_sum`: Sum of eigenvector components
- `meaning`: Semantic meaning
- `usage`: Usage description
- `labels`: Classification labels
## Natural Classes
- very_low: 25.19%
- low: 25.32%
- medium: 24.65%
- high: 20.40%
- very_high: 4.44%
## Correlations
- Gödel ↔ Shard: 1.000 (perfect!)
- Eigenvector sum ↔ Gödel: 0.996
## Usage
```python
from datasets import load_dataset
ds = load_dataset("introspector/zkprologml")
print(ds['train'][0])
```
## Links
- Dashboard: https://huggingface.co/spaces/introspector/zkprologml
- GitHub: https://github.com/meta-introspector/zkprologml