Added links to paper

README.md

@@ -11,7 +11,7 @@ language:
 ## Pretrained juridical BERT model for Romanian 
 
 BERT Romanian juridical model trained using a masked language modeling (MLM) and next sentence prediction (NSP) objective. 
-It was introduced in this [paper](TODO). Two BERT models were released: **jurBERT-base** and **jurBERT-large**, all versions uncased.
+It was introduced in this [paper](https://aclanthology.org/2021.nllp-1.8/). Two BERT models were released: **jurBERT-base** and **jurBERT-large**, all versions uncased.
 
 | Model          | Weights   |   L    |   H    |    A   | MLM accuracy   | NSP accuracy   |
 |----------------|:---------:|:------:|:------:|:------:|:--------------:|:--------------:|
@@ -98,7 +98,7 @@ We report Mean AUC and Std AUC on the task of predicting the outcome of a case.
 | *jurBERT-base*      | *59.65*  | *1.16*   |
 | *jurBERT-base + hf* | **61.46**| **1.76** |
 
-For complete results and discussion please refer to the [paper](TODO).
+For complete results and discussion please refer to the [paper](https://aclanthology.org/2021.nllp-1.8/).
 
 ### BibTeX entry and citation info