model card
Browse files
README.md
ADDED
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- mt
|
4 |
+
datasets:
|
5 |
+
- MLRS/korpus_malti
|
6 |
+
model-index:
|
7 |
+
- name: BERTu
|
8 |
+
results:
|
9 |
+
- task:
|
10 |
+
type: dependency-parsing
|
11 |
+
name: Dependency Parsing
|
12 |
+
dataset:
|
13 |
+
type: universal_dependencies
|
14 |
+
args: mt_mudt
|
15 |
+
name: Maltese Universal Dependencies Treebank (MUDT)
|
16 |
+
metrics:
|
17 |
+
- type: uas
|
18 |
+
value: 92.31
|
19 |
+
name: Unlabelled Attachment Score
|
20 |
+
- type: las
|
21 |
+
value: 88.14
|
22 |
+
name: Labelled Attachment Score
|
23 |
+
- task:
|
24 |
+
type: part-of-speech-tagging
|
25 |
+
name: Part-of-Speech Tagging
|
26 |
+
dataset:
|
27 |
+
type: mlrs_pos
|
28 |
+
name: MLRS POS dataset
|
29 |
+
metrics:
|
30 |
+
- type: accuracy
|
31 |
+
value: 98.58
|
32 |
+
name: UPOS Accuracy
|
33 |
+
args: upos
|
34 |
+
- type: accuracy
|
35 |
+
value: 98.54
|
36 |
+
name: XPOS Accuracy
|
37 |
+
args: xpos
|
38 |
+
- task:
|
39 |
+
type: named-entity-recognition
|
40 |
+
name: Named Entity Recognition
|
41 |
+
dataset:
|
42 |
+
type: wikiann
|
43 |
+
name: WikiAnn (Maltese)
|
44 |
+
args: mt
|
45 |
+
metrics:
|
46 |
+
- type: f1
|
47 |
+
args: span
|
48 |
+
value: 86.77
|
49 |
+
name: Span-based F1
|
50 |
+
- task:
|
51 |
+
type: sentiment-analysis
|
52 |
+
name: Sentiment Analysis
|
53 |
+
dataset:
|
54 |
+
type: mt-sentiment-analysis
|
55 |
+
name: Maltese Sentiment Analysis Dataset
|
56 |
+
metrics:
|
57 |
+
- type: f1
|
58 |
+
args: macro
|
59 |
+
value: 78.96
|
60 |
+
name: Macro-averaged F1
|
61 |
+
widget:
|
62 |
+
- text: "Malta hija gżira fil-[MASK]."
|
63 |
+
---
|
64 |
+
|
65 |
+
# BERTu
|
66 |
+
|
67 |
+
A Maltese monolingual model pre-trained from scratch on the Korpus Malti v4.0 using the BERT (base) architecture.
|