Commit ·
7f410c5
1
Parent(s): cd9a24b
Update README.md
Browse files
README.md
CHANGED
|
@@ -34,9 +34,11 @@ A train set is provided, which can be split into validation and test if needed.
|
|
| 34 |
|
| 35 |
### Data Fields
|
| 36 |
|
|
|
|
| 37 |
"contents": the text
|
| 38 |
"metadata": contains information about the source(s) of text that the text comes from. Multiple sources means that the example is concatenated from two sources.
|
| 39 |
"id": Ignore - a non-unique identifier
|
|
|
|
| 40 |
|
| 41 |
## Dataset Creation
|
| 42 |
We first select 102.4M examples then concatenate every two examples to create 51.2M examples.
|
|
|
|
| 34 |
|
| 35 |
### Data Fields
|
| 36 |
|
| 37 |
+
```
|
| 38 |
"contents": the text
|
| 39 |
"metadata": contains information about the source(s) of text that the text comes from. Multiple sources means that the example is concatenated from two sources.
|
| 40 |
"id": Ignore - a non-unique identifier
|
| 41 |
+
```
|
| 42 |
|
| 43 |
## Dataset Creation
|
| 44 |
We first select 102.4M examples then concatenate every two examples to create 51.2M examples.
|