German 1.0.0 added

README.md

@@ -19,8 +19,9 @@ training data, measured results and known limits on each card.
 | [gector-bert-it](https://huggingface.co/proofbird/gector-bert-it) | Italian | 1.0.0 | 512 fixed · 0 made worse | 0.10% |
 | [gector-bert-es](https://huggingface.co/proofbird/gector-bert-es) | Spanish | 1.2.0 | 430 fixed · 0 made worse | 0.50% |
 | [gector-bert-fr](https://huggingface.co/proofbird/gector-bert-fr) | French | 1.1.0 | 472 fixed · 0 made worse | 0.38% |
+| [gector-bert-de](https://huggingface.co/proofbird/gector-bert-de) | German | 1.0.0 | 168 fixed · 0 made worse | 0.53% |
 
-German follows. Every model card has a quick start: one Python file, or one click to run it in Google Colab.
+Every model card has a quick start: one Python file, or one click to run it in Google Colab.
 
 **How they're made:** synthetic errors over Wikipedia and Tatoeba sentences — no learner essays, no non-commercial training
 data, no user text. Precision first: a missed error is acceptable, a wrong correction is not.