Explain JSON metadata parsing
Browse files
README.md
CHANGED
|
@@ -139,9 +139,11 @@ Some records also include counts of replacements for email addresses, IP address
|
|
| 139 |
|
| 140 |
The full dataset spans several terabytes, so streaming is recommended:
|
| 141 |
|
|
|
|
|
|
|
| 142 |
```python
|
| 143 |
-
from datasets import load_dataset
|
| 144 |
import json
|
|
|
|
| 145 |
|
| 146 |
dataset = load_dataset(
|
| 147 |
"openbmb/Ultra-FineWeb-L1",
|
|
|
|
| 139 |
|
| 140 |
The full dataset spans several terabytes, so streaming is recommended:
|
| 141 |
|
| 142 |
+
The `meta` field is stored as a JSON-encoded string. Use `json.loads` to access its metadata fields; the `content` field is already plain text.
|
| 143 |
+
|
| 144 |
```python
|
|
|
|
| 145 |
import json
|
| 146 |
+
from datasets import load_dataset
|
| 147 |
|
| 148 |
dataset = load_dataset(
|
| 149 |
"openbmb/Ultra-FineWeb-L1",
|