vibhorag101
commited on
Commit
•
d95fd9a
1
Parent(s):
c6ef525
Update README.md
Browse files
README.md
CHANGED
@@ -42,4 +42,8 @@ size_categories:
|
|
42 |
- Removed any consecutive characters repeated more than 3 times.
|
43 |
- Tokenised the text, then lemmatized it and then removed the stopwords (excluding not).
|
44 |
- The `class_label` column was renamed to `label` for use with trainer API.
|
45 |
-
- The evaluation set had ~23000 samples, while the training set had ~186k samples, i.e. a 80:10:10 (train:test:val) split.
|
|
|
|
|
|
|
|
|
|
42 |
- Removed any consecutive characters repeated more than 3 times.
|
43 |
- Tokenised the text, then lemmatized it and then removed the stopwords (excluding not).
|
44 |
- The `class_label` column was renamed to `label` for use with trainer API.
|
45 |
+
- The evaluation set had ~23000 samples, while the training set had ~186k samples, i.e. a 80:10:10 (train:test:val) split.
|
46 |
+
|
47 |
+
### Note
|
48 |
+
Since this dataset was preprocessed, and stopwords and punctuation marks such as "?!" were removed from it, it might be possible that in some cases
|
49 |
+
that, the text is having incorrect labels, as the meaning changed against the original text after preprocessing.
|