z-dickson/CAP_coded_US_Congressional_bills

This model predicts the issue category of US Congressional bills.

The model is trained on ~250k US Congressional bills from 1950-2015.

The issue coding scheme follows the Comparative Agenda Project: https://www.comparativeagendas.net/pages/master-codebook

The model is cased (case sensitive)

Any questions on the model and training data feel free to message me on twitter - @sachary_

Train Loss: 0.1318; Train Sparse Categorical Accuracy: 0.9268; Validation Loss: 0.2439; Validation Sparse Categorical Accuracy: 0.9161

The following hyperparameters were used during training:

optimizer: {'name': 'Adam', 'learning_rate': 5e-05, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False} training_precision: float32

Training hyperparameters

Framework versions

• Transformers 4.19.3
• TensorFlow 2.8.2
• Tokenizers 0.12.1