Update README.md

README.md

@@ -7,7 +7,7 @@ language:
 ---
 # XLM-R-BERTić
 
-This model was produced by pre-training [XLM-Roberta-large](https://huggingface.co/xlm-roberta-large) 48k steps on South Slavic languages.
+This model was produced by pre-training [XLM-Roberta-large](https://huggingface.co/xlm-roberta-large) 48k steps on South Slavic languages using [XLM-R-BERTić dataset](https://huggingface.co/datasets/classla/xlm-r-bertic-data)
 
 # Benchmarking
 Three tasks were chosen for model evaluation:
@@ -24,7 +24,7 @@ Mean F1 scores were used to evaluate performance.
 
 | system                                                                 | dataset | F1 score |
 |:-----------------------------------------------------------------------|:--------|---------:|
-| **XLM-R-BERTić**  (this model)                                         | hr500k  |    0.927 |
+| [XLM-R-BERTić](https://huggingface.co/classla/xlm-r-bertic)            | hr500k  |    0.927 |
 | [BERTić](https://huggingface.co/classla/bcms-bertic)                   | hr500k  |    0.925 |
 | [XLM-R-SloBERTić](https://huggingface.co/classla/xlm-r-slobertic)      | hr500k  |    0.923 |
 | [XLM-Roberta-Large](https://huggingface.co/xlm-roberta-large)          | hr500k  |    0.919 |
@@ -34,7 +34,7 @@ Mean F1 scores were used to evaluate performance.
 | system                                                                 | dataset  | F1 score |
 |:-----------------------------------------------------------------------|:---------|---------:|
 | [XLM-R-SloBERTić](https://huggingface.co/classla/xlm-r-slobertic)      | ReLDI-hr |    0.812 |
-| **XLM-R-BERTić**  (this model)                                         | ReLDI-hr |    0.809 |
+| [XLM-R-BERTić](https://huggingface.co/classla/xlm-r-bertic)            | ReLDI-hr |    0.809 |
 | [crosloengual-bert](https://huggingface.co/EMBEDDIA/crosloengual-bert) | ReLDI-hr |    0.794 |
 | [BERTić](https://huggingface.co/classla/bcms-bertic)                   | ReLDI-hr |    0.792 |
 | [XLM-Roberta-Large](https://huggingface.co/xlm-roberta-large)          | ReLDI-hr |    0.791 |
@@ -43,7 +43,7 @@ Mean F1 scores were used to evaluate performance.
 | system                                                                 | dataset    | F1 score |
 |:-----------------------------------------------------------------------|:-----------|---------:|
 | [XLM-R-SloBERTić](https://huggingface.co/classla/xlm-r-slobertic)      | SETimes.SR |    0.949 |
-| **XLM-R-BERTić**   (this model)                                        | SETimes.SR |    0.940 |
+| [XLM-R-BERTić](https://huggingface.co/classla/xlm-r-bertic)            | SETimes.SR |    0.940 |
 | [BERTić](https://huggingface.co/classla/bcms-bertic)                   | SETimes.SR |    0.936 |
 | [XLM-Roberta-Large](https://huggingface.co/xlm-roberta-large)          | SETimes.SR |    0.933 |
 | [crosloengual-bert](https://huggingface.co/EMBEDDIA/crosloengual-bert) | SETimes.SR |    0.922 |
@@ -51,7 +51,7 @@ Mean F1 scores were used to evaluate performance.
 
 | system                                                                 | dataset  | F1 score |
 |:-----------------------------------------------------------------------|:---------|---------:|
-| **XLM-R-BERTić** (this model)                                          | ReLDI-sr |    0.841 |
+| [XLM-R-BERTić](https://huggingface.co/classla/xlm-r-bertic)            | ReLDI-sr |    0.841 |
 | [XLM-R-SloBERTić](https://huggingface.co/classla/xlm-r-slobertic)      | ReLDI-sr |    0.824 |
 | [BERTić](https://huggingface.co/classla/bcms-bertic)                   | ReLDI-sr |    0.798 |
 | [XLM-Roberta-Large](https://huggingface.co/xlm-roberta-large)          | ReLDI-sr |    0.774 |
@@ -69,7 +69,7 @@ The procedure is explained in greater detail in the dedicated [benchmarking repo
 | [BERTić](https://huggingface.co/classla/bcms-bertic)                   | ParlaSent_BCS.jsonl | ParlaSent_BCS_test.jsonl | 0.612 |
 | [XLM-R-SloBERTić](https://huggingface.co/classla/xlm-r-slobertic)      | ParlaSent_BCS.jsonl | ParlaSent_BCS_test.jsonl | 0.607 |
 | [XLM-Roberta-Large](https://huggingface.co/xlm-roberta-large)          | ParlaSent_BCS.jsonl | ParlaSent_BCS_test.jsonl | 0.605 |
-| **XLM-R-BERTić**    (this model)                                       | ParlaSent_BCS.jsonl | ParlaSent_BCS_test.jsonl | 0.601 |
+| [XLM-R-BERTić](https://huggingface.co/classla/xlm-r-bertic)            | ParlaSent_BCS.jsonl | ParlaSent_BCS_test.jsonl | 0.601 |
 | [crosloengual-bert](https://huggingface.co/EMBEDDIA/crosloengual-bert) | ParlaSent_BCS.jsonl | ParlaSent_BCS_test.jsonl | 0.537 |
 | [XLM-Roberta-Base](https://huggingface.co/xlm-roberta-base)            | ParlaSent_BCS.jsonl | ParlaSent_BCS_test.jsonl | 0.500 |
 | dummy (mean)                                                           | ParlaSent_BCS.jsonl | ParlaSent_BCS_test.jsonl | -0.12 |
@@ -77,12 +77,13 @@ The procedure is explained in greater detail in the dedicated [benchmarking repo
 
 ## COPA
 
+Two South Slavic COPA datasets were used, [COPA-HR](https://huggingface.co/datasets/classla/copa_hr) and [COPA-SR_lat](https://huggingface.co/datasets/classla/COPA-SR_lat).
 
 | system                                                                 | dataset | Accuracy score |
 |:-----------------------------------------------------------------------|:--------|---------------:|
 | [BERTić](https://huggingface.co/classla/bcms-bertic)                   | Copa-SR |          0.689 |
 | [XLM-R-SloBERTić](https://huggingface.co/classla/xlm-r-slobertic)      | Copa-SR |          0.665 |
-| **XLM-R-BERTić**                (this model)                           | Copa-SR |          0.637 |
+| [XLM-R-BERTić](https://huggingface.co/classla/xlm-r-bertic)            | Copa-SR |          0.637 |
 | [crosloengual-bert](https://huggingface.co/EMBEDDIA/crosloengual-bert) | Copa-SR |          0.607 |
 | [XLM-Roberta-Base](https://huggingface.co/xlm-roberta-base)            | Copa-SR |          0.573 |
 | [XLM-Roberta-Large](https://huggingface.co/xlm-roberta-large)          | Copa-SR |          0.570 |
@@ -92,13 +93,11 @@ The procedure is explained in greater detail in the dedicated [benchmarking repo
 |:-----------------------------------------------------------------------|:--------|---------------:|
 | [BERTić](https://huggingface.co/classla/bcms-bertic)                   | Copa-HR |          0.669 |
 | [XLM-R-SloBERTić](https://huggingface.co/classla/xlm-r-slobertic)      | Copa-HR |          0.628 |
-| **XLM-R-BERTić**                            (this model)               | Copa-HR |          0.635 |
+| [XLM-R-BERTić](https://huggingface.co/classla/xlm-r-bertic)            | Copa-HR |          0.635 |
 | [crosloengual-bert](https://huggingface.co/EMBEDDIA/crosloengual-bert) | Copa-HR |          0.669 |
 | [XLM-Roberta-Base](https://huggingface.co/xlm-roberta-base)            | Copa-HR |          0.585 |
 | [XLM-Roberta-Large](https://huggingface.co/xlm-roberta-large)          | Copa-HR |          0.571 |
 
-
-
 # Citation
 (to be added soon)
 # Authors