Spaces:
Paused
Paused
c4 - more data
Browse files
src/calibration_datasets.py
CHANGED
@@ -246,6 +246,15 @@ class C4Dataset(CalibrationDataset):
|
|
246 |
"en/c4-train.00008-of-01024.json.gz",
|
247 |
"en/c4-train.00009-of-01024.json.gz",
|
248 |
"en/c4-train.00010-of-01024.json.gz",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
249 |
],
|
250 |
},
|
251 |
"split": "train"
|
|
|
246 |
"en/c4-train.00008-of-01024.json.gz",
|
247 |
"en/c4-train.00009-of-01024.json.gz",
|
248 |
"en/c4-train.00010-of-01024.json.gz",
|
249 |
+
"en/c4-train.00011-of-01024.json.gz",
|
250 |
+
"en/c4-train.00012-of-01024.json.gz",
|
251 |
+
"en/c4-train.00013-of-01024.json.gz",
|
252 |
+
"en/c4-train.00014-of-01024.json.gz",
|
253 |
+
"en/c4-train.00015-of-01024.json.gz",
|
254 |
+
"en/c4-train.00016-of-01024.json.gz",
|
255 |
+
"en/c4-train.00017-of-01024.json.gz",
|
256 |
+
"en/c4-train.00018-of-01024.json.gz",
|
257 |
+
"en/c4-train.00019-of-01024.json.gz",
|
258 |
],
|
259 |
},
|
260 |
"split": "train"
|