parl-german-ner-bt

Annotation Layer: NER

This model is part of GePaDeU, which equips parliamentary debates of the German Bundestag with rich semantic and pragmatic information across multiple annotation layers.

parl-german-ner-bt is trained on German parliamentary speeches to tag a sequence with fine-grained named entities (e.g., geo-political entities, persons, organizations). The NER tag inventory is adapted from the OntoNotes NER inventory.

See schlenker/parl-german-ner which is trained on a mix of news data and parliamentary speeches.

---

🔍 Model Overview

• Task Type: Token classification
• Base Model: GBERT large
• Fine-tuning method: full fine-tuning
• Language: German

---

📚 Dataset

Models were trained and evaluated on 267 manually annotated parliamentary speeches (resulting in 10,564 annotated entities).

Data Splits

• Train: 3,020 instances
• Dev: 405 instances
• Test: 1,409 instances

---

🏋️ Model Training

---

📊 Evaluation

---

🚀 How to Use

Please, refer to our GitHub repo for detailed instructions on the required input format and how to run the model.

---

⚠️ Limitations