Datasets:

Modalities:
Text
Formats:
parquet
ArXiv:
Libraries:
Datasets
Dask
jqhoogland commited on
Commit
a5d6a32
·
verified ·
1 Parent(s): c70f82d

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +22 -0
README.md CHANGED
@@ -19,3 +19,25 @@ configs:
19
  - split: train
20
  path: data/train-*
21
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
19
  - split: train
20
  path: data/train-*
21
  ---
22
+
23
+ ## Dataset Creation Process
24
+ These subsets were created by streaming over the rows from [`monology/pile-uncopyrighted`](https://huggingface.co/datasets/monology/pile-uncopyrighted) and filtering by the meta column. Each subset is generally limited to the first 100,000 qualifying rows encountered.
25
+
26
+ ## Citations
27
+
28
+ If you use this dataset, please cite the original Pile papers:
29
+
30
+ ```bibtex
31
+ @article{gao2020pile,
32
+ title={The Pile: An 800GB dataset of diverse text for language modeling},
33
+ author={Gao, Leo and Biderman, Stella and Black, Sid and Golding, Laurence and Hoppe, Travis and Foster, Charles and Phang, Jason and He, Horace and Thite, Anish and Nabeshima, Noa and others},
34
+ journal={arXiv preprint arXiv:2101.00027},
35
+ year={2020}
36
+ }
37
+ @article{biderman2022datasheet,
38
+ title={Datasheet for the pile},
39
+ author={Biderman, Stella and Bicheno, Kieran and Gao, Leo},
40
+ journal={arXiv preprint arXiv:2201.07311},
41
+ year={2022}
42
+ }
43
+ ```