Update README.md
Browse files
README.md
CHANGED
@@ -22,7 +22,8 @@ dataset_size: 26.1 GB
|
|
22 |
This dataset combines [wikipedia20220301.en](https://huggingface.co/datasets/wikipedia) and [bookcorpusopen](https://huggingface.co/datasets/bookcorpusopen),
|
23 |
and splits the data into smaller chunks, of size ~820 chars
|
24 |
(such that each item will be at least ~128 tokens for the average tokenizer).
|
25 |
-
The order of the items has been shuffled
|
|
|
26 |
The logic only splits on spaces, so the chunks are likely to be slightly larger than 820 chars.
|
27 |
The dataset has been normalized into lower case, with accents and non-english characters removed.
|
28 |
Items with less than 200 chars or more than 1000 chars have been removed.
|
|
|
22 |
This dataset combines [wikipedia20220301.en](https://huggingface.co/datasets/wikipedia) and [bookcorpusopen](https://huggingface.co/datasets/bookcorpusopen),
|
23 |
and splits the data into smaller chunks, of size ~820 chars
|
24 |
(such that each item will be at least ~128 tokens for the average tokenizer).
|
25 |
+
The order of the items in this dataset has been shuffled,
|
26 |
+
meaning it's faster to iterate over than using `dataset.shuffle`.
|
27 |
The logic only splits on spaces, so the chunks are likely to be slightly larger than 820 chars.
|
28 |
The dataset has been normalized into lower case, with accents and non-english characters removed.
|
29 |
Items with less than 200 chars or more than 1000 chars have been removed.
|