lvwerra HF staff commited on
Commit
adf0e46
1 Parent(s): e7f3408

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +26 -0
README.md ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - python
4
+ - code
5
+ ---
6
+
7
+ # CodeParrot 🦜 Dataset Cleaned
8
+
9
+ ## What is it?
10
+
11
+ A dataset of Python files from Github. This is the deduplicated version of the [codeparrot](https://huggingface.co/datasets/transformersbook/codeparrot).
12
+
13
+ ## Processing
14
+ The original dataset contains a lot of duplicated and noisy data. Therefore, the dataset was cleaned with the following steps:
15
+ - Deduplication
16
+ - Remove exact matches
17
+ - Filtering
18
+ - Average line length < 100
19
+ - Maximum line length < 1000
20
+ - Alpha numeric characters fraction > 0.25
21
+ - Remove auto-generated files (keyword search)
22
+
23
+ For more details see the preprocessing script in the transformers repository [here](https://github.com/huggingface/transformers/tree/master/examples/research_projects/codeparrot).
24
+
25
+ ## Splits
26
+ The dataset is split in a [train](https://huggingface.co/datasets/lvwerra/codeparrot-clean-train) and [validation](https://huggingface.co/datasets/lvwerra/codeparrot-clean-valid) split used for training and evaluation.