Instructions to use Sennodipoi/LayoutLMv3-kleisterNDA with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Sennodipoi/LayoutLMv3-kleisterNDA with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("token-classification", model="Sennodipoi/LayoutLMv3-kleisterNDA")# Load model directly from transformers import AutoProcessor, AutoModelForTokenClassification processor = AutoProcessor.from_pretrained("Sennodipoi/LayoutLMv3-kleisterNDA") model = AutoModelForTokenClassification.from_pretrained("Sennodipoi/LayoutLMv3-kleisterNDA") - Notebooks
- Google Colab
- Kaggle
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
LayoutLMv3 fine-tuned on the Kleister-NDA dataset. Code (including pre-processing) and results are available at the official GitHub repository of my Master Degree thesis .
Results obtained with seqeval in strict mode:
| Precision | Recall | F1-score | Variance (F1) | |
|---|---|---|---|---|
| EFFECTIVE_DATE | 0.92 | 0.99 | 0.95 | 5e-5 |
| JURISDICTION | 0.87 | 0.88 | 0.88 | 8e-6 |
| PARTY | 0.92 | 0.99 | 0.95 | 5e-5 |
| TERM | 1 | 1 | 1 | 0 |
| Micro avg | 0.91 | 0.96 | 0.94 | 2e-5 |
| Macro avg | 0.92 | 0.96 | 0.94 | 3e-7 |
| Weighted avg | 0.91 | 0.96 | 0.94 | 2e-5 |
Since I used the same segmentation strategy of the original paper i.e. using the labels to create segments, the scores are not directly comparable with the other LayoutLM versions.
- Downloads last month
- 7