redone yaml following GLUE

README.md

@@ -1,74 +1,154 @@
 ---
-configs:
-- config_name: fake_news
-  data_files:
-  - split: train
+language:
+- en
+license:
+- mit
+size_categories:
+- 10K<n<100K
+multilinguality:
+- monolingual
+task_categories:
+- text-classification
+pretty_name: GDDs (Generalized Deception Dataset)
+tags:
+- deception-detection
+- phishing
+- fake-news
+- opinion-spam
+- domain-adaptation
+dataset_info:
+- config: fake_news
+  features:
+  - name: text
+    dtype: string
+  - name: label
+    dtype:
+      class_label:
+        names:
+          '0': non-deceptive
+          '1': deceptive
+  splits:
+  - name: train
     path: fake_news/train.jsonl
-  - split: test
+  - name: test
     path: fake_news/test.jsonl
-  - split: validation
+  - name: validation
     path: fake_news/validation.jsonl
-- config_name: job_scams
-  data_files:
-  - split: train
+- config: job_scams
+  features:
+  - name: text
+    dtype: string
+  - name: label
+    dtype:
+      class_label:
+        names:
+          '0': non-deceptive
+          '1': deceptive
+  splits:
+  - name: train
     path: job_scams/train.jsonl
-  - split: test
+  - name: test
     path: job_scams/test.jsonl
-  - split: validation
+  - name: validation
     path: job_scams/validation.jsonl
-- config_name: phishing
-  data_files:
-  - split: train
+- config: phishing
+  features:
+  - name: text
+    dtype: string
+  - name: label
+    dtype:
+      class_label:
+        names:
+          '0': non-deceptive
+          '1': deceptive
+  splits:
+  - name: train
     path: phishing/train.jsonl
-  - split: test
+  - name: test
     path: phishing/test.jsonl
-  - split: validation
+  - name: validation
     path: phishing/validation.jsonl
-- config_name: political_statements
-  data_files:
-  - split: train
+- config: political_statements
+  features:
+  - name: text
+    dtype: string
+  - name: label
+    dtype:
+      class_label:
+        names:
+          '0': non-deceptive
+          '1': deceptive
+  splits:
+  - name: train
     path: political_statements/train.jsonl
-  - split: test
+  - name: test
     path: political_statements/test.jsonl
-  - split: validation
+  - name: validation
     path: political_statements/validation.jsonl
-- config_name: product_reviews
+- config: product_reviews
+  features:
+  - name: text
+    dtype: string
+  - name: label
+    dtype:
+      class_label:
+        names:
+          '0': non-deceptive
+          '1': deceptive
   data_files:
-  - split: train
+  - name: train
     path: product_reviews/train.jsonl
-  - split: test
+  - name: test
     path: product_reviews/test.jsonl
-  - split: validation
+  - name: validation
     path: product_reviews/validation.jsonl
-- config_name: sms
-  data_files:
-  - split: train
+- config: sms
+  features:
+  - name: text
+    dtype: string
+  - name: label
+    dtype:
+      class_label:
+        names:
+          '0': non-deceptive
+          '1': deceptive
+  splits:
+  - name: train
     path: sms/train.jsonl
-  - split: test
+  - name: test
     path: sms/test.jsonl
-  - split: validation
+  - name: validation
     path: sms/validation.jsonl
-- config_name: twitter_rumours
-  data_files:
-  - split: train
+- config: twitter_rumours
+  features:
+  - name: text
+    dtype: string
+  - name: label
+    dtype:
+      class_label:
+        names:
+          '0': non-deceptive
+          '1': deceptive
+  splits:
+  - name: train
     path: twitter_rumours/train.jsonl
-  - split: test
+  - name: test
     path: twitter_rumours/test.jsonl
-  - split: validation
+  - name: validation
     path: twitter_rumours/validation.jsonl
-license: mit
-task_categories:
-- text-classification
-language:
-- en
-size_categories:
-- 10K<n<100K
+config_names:
+- fake_news
+- job_scams
+- phishing
+- political_statements
+- product_reviews
+- sms
+- twitter_rumours
 ---
 
 # GDDs-2.0
 
-The Generalized Deception Dataset version 2.0 is a labeled corpus containing over 95000 samples of
-deceptive and truthful texts from a number of independent domains and tasks. 
+The Generalized Deception Dataset version 2.0 is a labeled corpus containing over 95000 samples of deceptive and truthful texts from a number of independent domains and tasks. 
 
 ## Authors
 
@@ -342,4 +422,4 @@ https://figshare.com/articles/dataset/PHEME_dataset_of_rumours_and_non-rumours/4
 was used in creation of this dataset. We took source tweets only, and ignored replies to them. 
 We used source tweet's label as being a rumour or non-rumour to label it as deceptive or non-deceptive.
 
-The training set contains 4631 samples, the validation and the test sets have 579 samples each.
+The training set contains 4631 samples, the validation and the test sets have 579 samples each