ner-german
| Property | Value |
|---|---|
| Downloads | 33,760 |
| Framework | PyTorch + Flair |
| Training Data | CoNLL-03 German |
| Performance | 87.94% F1-Score |
What is ner-german?
ner-german is a state-of-the-art Named Entity Recognition model specifically designed for German text processing. Built using the Flair framework, it employs contextual string embeddings combined with an LSTM-CRF architecture to identify and classify named entities in German text. The model recognizes four distinct entity types: Person (PER), Location (LOC), Organization (ORG), and Miscellaneous (MISC).
Implementation Details
The model utilizes a sophisticated architecture combining multiple embedding layers: German GloVe embeddings and bidirectional Flair embeddings (forward and backward). The implementation uses a hidden size of 256 units in its LSTM layer and was trained for 150 epochs on the CoNLL-03 German dataset.
- Stacked embeddings combining word-level and contextual representations
- LSTM-CRF sequence labeling architecture
- Trained on the revised CoNLL-03 German dataset
- Optimized for production use with easy integration via Flair library
Core Capabilities
- Accurate identification of person names with high confidence scores
- Precise detection of geographical locations
- Recognition of organization names in German text
- Classification of miscellaneous named entities
- Support for processing both single sentences and longer texts
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its use of stacked embeddings combining both traditional word embeddings and contextual string embeddings, specifically optimized for German language processing. Its high F1-score of 87.94% on the CoNLL-03 German dataset makes it particularly reliable for production use.
Q: What are the recommended use cases?
The model is ideal for applications requiring German named entity extraction, such as information extraction systems, content analysis tools, and automated document processing. It's particularly effective for identifying proper names, locations, and organizations in German text.
