Update README.md
Browse files
README.md
CHANGED
|
@@ -6,9 +6,6 @@ language:
|
|
| 6 |
- cnr
|
| 7 |
- hbs
|
| 8 |
|
| 9 |
-
tags:
|
| 10 |
-
- fill-mask
|
| 11 |
-
|
| 12 |
license: apache-2.0
|
| 13 |
---
|
| 14 |
|
|
@@ -24,10 +21,10 @@ Evaluation metric is (seqeval) microF1. Reported are means of five runs. Best re
|
|
| 24 |
|
| 25 |
Dataset | Language | Variety | CLASSLA | mBERT | cseBERT | BERTić
|
| 26 |
---|---|---|---|---|---|---
|
| 27 |
-
hr500k | Croatian | standard | 93.87 | 94.60 | 95.74 |
|
| 28 |
-
reldi-hr | Croatian | internet non-standard | - | 88.87 | 91.63 |
|
| 29 |
SETimes.SR | Serbian | standard | 95.00 | 95.50 | **96.41** | 96.31
|
| 30 |
-
reldi-sr | Serbian | internet non-standard | - | 91.26 | 93.54 |
|
| 31 |
|
| 32 |
## Named entity recognition
|
| 33 |
|
|
@@ -35,10 +32,10 @@ Evaluation metric is (seqeval) microF1. Reported are means of five runs. Best re
|
|
| 35 |
|
| 36 |
Dataset | Language | Variety | CLASSLA | mBERT | cseBERT | BERTić
|
| 37 |
---|---|---|---|---|---|---
|
| 38 |
-
hr500k | Croatian | standard | 80.13 | 85.67 | 88.98 |
|
| 39 |
-
reldi-hr | Croatian | internet non-standard | - | 76.06 | 81.38 |
|
| 40 |
SETimes.SR | Serbian | standard | 84.64 | **92.41** | 92.28 | 92.02
|
| 41 |
-
reldi-sr | Serbian | internet non-standard | - | 81.29 | 82.76 |
|
| 42 |
|
| 43 |
|
| 44 |
## Geolocation prediction
|
|
@@ -61,5 +58,5 @@ System | Accuracy
|
|
| 61 |
random | 50.00
|
| 62 |
mBERT | 54.12
|
| 63 |
cseBERT | 61.80
|
| 64 |
-
BERTić |
|
| 65 |
|
|
|
|
| 6 |
- cnr
|
| 7 |
- hbs
|
| 8 |
|
|
|
|
|
|
|
|
|
|
| 9 |
license: apache-2.0
|
| 10 |
---
|
| 11 |
|
|
|
|
| 21 |
|
| 22 |
Dataset | Language | Variety | CLASSLA | mBERT | cseBERT | BERTić
|
| 23 |
---|---|---|---|---|---|---
|
| 24 |
+
hr500k | Croatian | standard | 93.87 | 94.60 | 95.74 | **95.81*****
|
| 25 |
+
reldi-hr | Croatian | internet non-standard | - | 88.87 | 91.63 | **92.28*****
|
| 26 |
SETimes.SR | Serbian | standard | 95.00 | 95.50 | **96.41** | 96.31
|
| 27 |
+
reldi-sr | Serbian | internet non-standard | - | 91.26 | 93.54 | **93.90*****
|
| 28 |
|
| 29 |
## Named entity recognition
|
| 30 |
|
|
|
|
| 32 |
|
| 33 |
Dataset | Language | Variety | CLASSLA | mBERT | cseBERT | BERTić
|
| 34 |
---|---|---|---|---|---|---
|
| 35 |
+
hr500k | Croatian | standard | 80.13 | 85.67 | 88.98 | **89.21******
|
| 36 |
+
reldi-hr | Croatian | internet non-standard | - | 76.06 | 81.38 | **83.05******
|
| 37 |
SETimes.SR | Serbian | standard | 84.64 | **92.41** | 92.28 | 92.02
|
| 38 |
+
reldi-sr | Serbian | internet non-standard | - | 81.29 | 82.76 | **87.92******
|
| 39 |
|
| 40 |
|
| 41 |
## Geolocation prediction
|
|
|
|
| 58 |
random | 50.00
|
| 59 |
mBERT | 54.12
|
| 60 |
cseBERT | 61.80
|
| 61 |
+
BERTić | **65.76****
|
| 62 |
|