sm-subgroup-classifier / en_NA-nb-OP /training_details.txt
erikhenriksson's picture
Upload folder using huggingface_hub
f71be9a verified
raw
history blame contribute delete
940 Bytes
Training Details for en_NA-nb-OP
========================================
Language: en
Register: NA-nb-OP
Training Date: 2025-09-26 14:08:46
Data Summary:
- Total samples: 1536
- Training samples: 1228
- Test samples: 308
- Embedding dimension: 1024
Classes:
- Number of classes: 4
- Class names: '', 'culture', 'dining', 'lifestyle'
- Class distribution: {'': 747, 'culture': 327, 'dining': 172, 'lifestyle': 290}
Cross-Validation Results:
- CV folds: 5
- CV scores: [0.9634146341463414, 0.959349593495935, 0.9634146341463414, 0.9877551020408163, 0.9755102040816327]
- CV mean: 0.9699
- CV std: 0.0104
- CV confidence interval: 0.9699 ± 0.0209
Final Performance:
- Test accuracy: 0.9643
Model Configuration:
- Algorithm: Logistic Regression
- Regularization (C): 1.0
- Feature scaling: StandardScaler
- Random state: 42
Files:
- Classifier: model.pkl
- Scaler: scaler.pkl
- Metadata: metadata.pkl
- This file: training_details.txt