Frame Finder

🚀 Frame Classification

This model is designed for frame classification, trained on FrameNet v1.7, aiming to solve the problem of accurately classifying frames in text and providing valuable insights for natural language processing tasks.

🚀 Quick Start

This model is trained on FrameNet v1.7. You can check out the training dataset here.

The data is loaded using the following code:

ds = dataset.load_dataset('liyucheng/FrameNet_v17', name = 'frame_label')

This approach flattens all frame annotations to specific sentences, transforming frame classification into a sequence tagging task.

📚 Documentation

Metrics

The performance metrics of the model are as follows:

{'accuracy_score': 0.8382018348623853, 'precision': 0.8382018348623853, 'recall': 0.8382018348623853, 'micro_f1': 0.8382018348623853, 'macro_f1': 0.45824850358482677}

📄 License

The project is licensed under the CC license.