Mariusz Kossakowski commited on
Commit
90966f7
1 Parent(s): c28d81d
clarin_datasets/cst_wikinews_dataset.py CHANGED
@@ -14,7 +14,10 @@ class CSTWikinewsDataset(DatasetToShow):
14
  """
15
 
16
  def load_data(self):
17
- DatasetToShow.load_data(self)
 
 
 
18
 
19
  def show_dataset(self):
20
  pass
 
14
  """
15
 
16
  def load_data(self):
17
+ raw_dataset = load_dataset(self.dataset_name)
18
+ self.data_dict = {
19
+ subset: raw_dataset[subset].to_pandas() for subset in self.subsets
20
+ }
21
 
22
  def show_dataset(self):
23
  pass
clarin_datasets/dataset_to_show.py CHANGED
@@ -13,11 +13,7 @@ class DatasetToShow(ABC):
13
 
14
  @abstractmethod
15
  def load_data(self):
16
- raw_dataset = load_dataset(self.dataset_name)
17
- self.data_dict = {
18
- subset: raw_dataset[subset].to_pandas()
19
- for subset in self.subsets
20
- }
21
 
22
  @abstractmethod
23
  def show_dataset(self):
 
13
 
14
  @abstractmethod
15
  def load_data(self):
16
+ pass
 
 
 
 
17
 
18
  @abstractmethod
19
  def show_dataset(self):
clarin_datasets/kpwr_ner_datasets.py CHANGED
@@ -5,7 +5,6 @@ from clarin_datasets.dataset_to_show import DatasetToShow
5
 
6
 
7
  class KpwrNerDataset(DatasetToShow):
8
-
9
  def __init__(self):
10
  DatasetToShow.__init__(self)
11
  self.dataset_name = "clarin-pl/kpwr-ner"
@@ -35,8 +34,7 @@ class KpwrNerDataset(DatasetToShow):
35
  def load_data(self):
36
  raw_dataset = load_dataset(self.dataset_name)
37
  self.data_dict = {
38
- subset: raw_dataset[subset].to_pandas()
39
- for subset in self.subsets
40
  }
41
 
42
  def show_dataset(self):
@@ -50,13 +48,3 @@ class KpwrNerDataset(DatasetToShow):
50
  with description:
51
  st.header("Dataset description")
52
  st.write(self.description)
53
-
54
-
55
-
56
-
57
-
58
-
59
-
60
-
61
-
62
-
 
5
 
6
 
7
  class KpwrNerDataset(DatasetToShow):
 
8
  def __init__(self):
9
  DatasetToShow.__init__(self)
10
  self.dataset_name = "clarin-pl/kpwr-ner"
 
34
  def load_data(self):
35
  raw_dataset = load_dataset(self.dataset_name)
36
  self.data_dict = {
37
+ subset: raw_dataset[subset].to_pandas() for subset in self.subsets
 
38
  }
39
 
40
  def show_dataset(self):
 
48
  with description:
49
  st.header("Dataset description")
50
  st.write(self.description)
 
 
 
 
 
 
 
 
 
 
clarin_datasets/nkjp_pos_dataset.py CHANGED
@@ -14,7 +14,10 @@ class NkjpPosDataset(DatasetToShow):
14
  """
15
 
16
  def load_data(self):
17
- DatasetToShow.load_data(self)
 
 
 
18
 
19
  def show_dataset(self):
20
  pass
 
14
  """
15
 
16
  def load_data(self):
17
+ raw_dataset = load_dataset(self.dataset_name)
18
+ self.data_dict = {
19
+ subset: raw_dataset[subset].to_pandas() for subset in self.subsets
20
+ }
21
 
22
  def show_dataset(self):
23
  pass
clarin_datasets/polemo_dataset.py CHANGED
@@ -24,7 +24,10 @@ class PolemoDataset(DatasetToShow):
24
  annotated with four labels: positive, negative, neutral, or ambiguous. """
25
 
26
  def load_data(self):
27
- DatasetToShow.show_dataset(self)
 
 
 
28
 
29
  def show_dataset(self):
30
  header = st.container()
 
24
  annotated with four labels: positive, negative, neutral, or ambiguous. """
25
 
26
  def load_data(self):
27
+ raw_dataset = load_dataset(self.dataset_name)
28
+ self.data_dict = {
29
+ subset: raw_dataset[subset].to_pandas() for subset in self.subsets
30
+ }
31
 
32
  def show_dataset(self):
33
  header = st.container()
clarin_datasets/punctuation_restoration_dataset.py CHANGED
@@ -37,7 +37,10 @@ class PunctuationRestorationDataset(DatasetToShow):
37
  """
38
 
39
  def load_data(self):
40
- DatasetToShow.load_data(self)
 
 
 
41
 
42
  def show_dataset(self):
43
  pass
 
37
  """
38
 
39
  def load_data(self):
40
+ raw_dataset = load_dataset(self.dataset_name)
41
+ self.data_dict = {
42
+ subset: raw_dataset[subset].to_pandas() for subset in self.subsets
43
+ }
44
 
45
  def show_dataset(self):
46
  pass