Datasets:
Tasks:
Text Classification
Modalities:
Text
Sub-tasks:
multi-class-classification
Languages:
English
Size:
10K - 100K
ArXiv:
License:
More bugfixes to get a working copy (squashed)
Browse files
raft.py
CHANGED
@@ -50,8 +50,8 @@ _LICENSE = ""
|
|
50 |
# This gets all folders within the directory named `data`
|
51 |
DATA_DIRS = next(os.walk('data'))[1]
|
52 |
|
53 |
-
_URLs = {s: {'train': f"
|
54 |
-
'test': f"
|
55 |
|
56 |
|
57 |
class Raft(datasets.GeneratorBasedBuilder):
|
@@ -73,18 +73,25 @@ class Raft(datasets.GeneratorBasedBuilder):
|
|
73 |
|
74 |
# TODO: Load task jsons
|
75 |
|
76 |
-
tasks =
|
77 |
for sd in DATA_DIRS:
|
78 |
with open(os.path.join('data', sd, 'task.json')) as f:
|
79 |
task_data = json.load(f)
|
80 |
-
tasks
|
81 |
|
82 |
-
BUILDER_CONFIGS = [
|
83 |
-
|
|
|
|
|
|
|
|
|
|
|
84 |
|
85 |
DEFAULT_CONFIG_NAME = "tai_safety_research" # It's not mandatory to have a default configuration. Just use one if it make sense.
|
86 |
|
87 |
def _info(self):
|
|
|
|
|
88 |
task = Raft.tasks[self.config.name]
|
89 |
# TODO: This method specifies the datasets.DatasetInfo object which contains informations and typings for the dataset
|
90 |
data_columns = {col_name: datasets.Value("string") for col_name in
|
@@ -92,7 +99,7 @@ class Raft(datasets.GeneratorBasedBuilder):
|
|
92 |
|
93 |
label_columns = {}
|
94 |
for label_name in task['label_columns']:
|
95 |
-
labels = task['label_columns'][label_name]
|
96 |
label_columns[label_name] = datasets.ClassLabel(len(labels), labels)
|
97 |
|
98 |
# Merge dicts
|
@@ -152,5 +159,5 @@ class Raft(datasets.GeneratorBasedBuilder):
|
|
152 |
if id_ == 0: # First row is column names
|
153 |
continue
|
154 |
if split == "test":
|
155 |
-
row += [""] * num_labels
|
156 |
-
|
|
|
50 |
# This gets all folders within the directory named `data`
|
51 |
DATA_DIRS = next(os.walk('data'))[1]
|
52 |
|
53 |
+
_URLs = {s: {'train': f"data/{s}/train.csv",
|
54 |
+
'test': f"data/{s}/test_unlabeled.csv"} for s in DATA_DIRS}
|
55 |
|
56 |
|
57 |
class Raft(datasets.GeneratorBasedBuilder):
|
|
|
73 |
|
74 |
# TODO: Load task jsons
|
75 |
|
76 |
+
tasks = {}
|
77 |
for sd in DATA_DIRS:
|
78 |
with open(os.path.join('data', sd, 'task.json')) as f:
|
79 |
task_data = json.load(f)
|
80 |
+
tasks[sd] = task_data
|
81 |
|
82 |
+
BUILDER_CONFIGS = []
|
83 |
+
for key in tasks:
|
84 |
+
td = tasks[key]
|
85 |
+
name = td['name']
|
86 |
+
description = td['description']
|
87 |
+
BUILDER_CONFIGS.append(datasets.BuilderConfig(name=name, version=VERSION,
|
88 |
+
description=description))
|
89 |
|
90 |
DEFAULT_CONFIG_NAME = "tai_safety_research" # It's not mandatory to have a default configuration. Just use one if it make sense.
|
91 |
|
92 |
def _info(self):
|
93 |
+
DEFAULT_LABEL_NAME = "Unlabeled"
|
94 |
+
|
95 |
task = Raft.tasks[self.config.name]
|
96 |
# TODO: This method specifies the datasets.DatasetInfo object which contains informations and typings for the dataset
|
97 |
data_columns = {col_name: datasets.Value("string") for col_name in
|
|
|
99 |
|
100 |
label_columns = {}
|
101 |
for label_name in task['label_columns']:
|
102 |
+
labels = ["Unlabeled"] + task['label_columns'][label_name]
|
103 |
label_columns[label_name] = datasets.ClassLabel(len(labels), labels)
|
104 |
|
105 |
# Merge dicts
|
|
|
159 |
if id_ == 0: # First row is column names
|
160 |
continue
|
161 |
if split == "test":
|
162 |
+
row += ["Unlabeled"] * num_labels
|
163 |
+
yield id_, {name: value for name, value in zip(column_names, row)}
|