Commit
•
7dc3145
1
Parent(s):
4aa689e
update description
Browse files
README.md
CHANGED
|
@@ -65,13 +65,9 @@ tags:
|
|
| 65 |
- nlp
|
| 66 |
---
|
| 67 |
|
|
|
|
| 68 |
|
| 69 |
-
|
| 70 |
-
All I know is that I looked everywhere: HuggingFace, research papers, GitHub, Kaggle, and Google search. I even fetched 20K+ tweets using the Twitter API.
|
| 71 |
-
Today (6/28/2023) I came across two newer HuggingFace datasets, so I remembered to credit them at the bottom of the page.
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
The deduplicated training data alone consists of 2,880,230 rows of comments and messages. Among these rows, 416,457 are classified as toxic, while the remaining 2,463,773 are considered neutral. Below is a table to illustrate the data composition:
|
| 75 |
| | Toxic | Neutral | Total |
|
| 76 |
|-------|----------|----------|----------|
|
| 77 |
| [multilingual-train-deduplicated.csv](./multilingual-train-deduplicated.csv) | 416,457 | 2,463,773 | 2,880,230 |
|
|
@@ -79,7 +75,7 @@ The deduplicated training data alone consists of 2,880,230 rows of comments and
|
|
| 79 |
| [multilingual-test.csv](./multilingual-test.csv) | 14,410 | 49,402 | 63,812 |
|
| 80 |
Each CSV file has two columns: `text` and `is_toxic`.
|
| 81 |
|
| 82 |
-
|
| 83 |
- Afrikaans
|
| 84 |
- Albanian
|
| 85 |
- Arabic
|
|
@@ -139,6 +135,12 @@ Each CSV file has two columns: `text` and `is_toxic`.
|
|
| 139 |
|
| 140 |
<br>
|
| 141 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 142 |
Known datasets:
|
| 143 |
- tomekkorbak/pile-toxicity-balanced2
|
| 144 |
- datasets/thai_toxicity_tweet
|
|
|
|
| 65 |
- nlp
|
| 66 |
---
|
| 67 |
|
| 68 |
+
This is a large multilingual toxicity dataset with nearly 3M rows of text data from 55 different languages.
|
| 69 |
|
| 70 |
+
The preprocessed training data alone consists of 2,880,230 rows of comments, tweets, and messages. Among these rows, 416,457 are classified as toxic, while the remaining 2,463,773 are considered neutral. Below is a table to illustrate the data composition:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 71 |
| | Toxic | Neutral | Total |
|
| 72 |
|-------|----------|----------|----------|
|
| 73 |
| [multilingual-train-deduplicated.csv](./multilingual-train-deduplicated.csv) | 416,457 | 2,463,773 | 2,880,230 |
|
|
|
|
| 75 |
| [multilingual-test.csv](./multilingual-test.csv) | 14,410 | 49,402 | 63,812 |
|
| 76 |
Each CSV file has two columns: `text` and `is_toxic`.
|
| 77 |
|
| 78 |
+
Supported languages:
|
| 79 |
- Afrikaans
|
| 80 |
- Albanian
|
| 81 |
- Arabic
|
|
|
|
| 135 |
|
| 136 |
<br>
|
| 137 |
|
| 138 |
+
|
| 139 |
+
### Original Source?
|
| 140 |
+
Around 11 months ago, I downloaded and preprocessed 2.7M rows of text data, but completely forgot the original source of these datasets...
|
| 141 |
+
All I know is that I looked everywhere: HuggingFace, research papers, GitHub, Kaggle, and Google search. I even fetched 20K+ tweets using the Twitter API.
|
| 142 |
+
Today (6/28/2023) I came across two newer HuggingFace datasets, so I remembered to credit them below.
|
| 143 |
+
|
| 144 |
Known datasets:
|
| 145 |
- tomekkorbak/pile-toxicity-balanced2
|
| 146 |
- datasets/thai_toxicity_tweet
|