readme: add initial version
Browse files
README.md
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
tags:
|
| 3 |
+
- flair
|
| 4 |
+
- token-classification
|
| 5 |
+
- sequence-tagger-model
|
| 6 |
+
language: de
|
| 7 |
+
inference: false
|
| 8 |
+
license: mit
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Towards Robust Named Entity Recognition for Historic German
|
| 12 |
+
|
| 13 |
+
Based on [our paper](https://www.aclweb.org/anthology/W19-4312/)
|
| 14 |
+
we release a new model trained on the LFT dataset.
|
| 15 |
+
|
| 16 |
+
**Note:** We use BPEmbeddings instead of the combination of
|
| 17 |
+
Wikipedia, Common Crawl and character embeddings (as used in the paper),
|
| 18 |
+
so save space and training/inferencing time.
|
| 19 |
+
|
| 20 |
+
# Results
|
| 21 |
+
|
| 22 |
+
| Dataset \ Run | Run 1 | Run 2 | Run 3† | Avg.
|
| 23 |
+
| ------------- | ----- | ----- | --------- | ------------
|
| 24 |
+
| Development | 76.32 | 76.13 | **76.36** | 76.27
|
| 25 |
+
| Test | 77.07 | 77.35 | 77.20 | 77.21
|
| 26 |
+
|
| 27 |
+
Paper reported an averaged F1-score of 77.51.
|
| 28 |
+
|
| 29 |
+
† denotes that this model is selected for upload.
|