stefan-it commited on
Commit
7c5f973
1 Parent(s): fb4dc6a

readme: add initial version

Browse files
Files changed (1) hide show
  1. README.md +34 -0
README.md ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language: de
3
+ license: mit
4
+ datasets:
5
+ - german-nlp-group/german_common_crawl
6
+ widget:
7
+ - text: "Heute ist ein [MASK] Tag"
8
+ ---
9
+
10
+ # GC4LM: A Colossal (Biased) language model for German
11
+ This repository presents a colossal (and biased) language model for German trained on the recently released
12
+ ["German colossal, clean Common Crawl corpus"](https://german-nlp-group.github.io/projects/gc4-corpus.html) (GC4),
13
+ with a total dataset size of ~844GB.
14
+
15
+ ---
16
+
17
+ **Disclaimer**: the presented and trained language models in this repository are for **research only** purposes.
18
+ The GC4 corpus - that was used for training - contains crawled texts from the internet. Thus, the language models can
19
+ be considered as highly biased, resulting in a model that encodes stereotypical associations along gender, race,
20
+ ethnicity and disability status. Before using and working with the released checkpoints, it is highly recommended
21
+ to read:
22
+
23
+ [On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?](https://faculty.washington.edu/ebender/papers/Stochastic_Parrots.pdf)
24
+
25
+ from Emily M. Bender, Timnit Gebru, Angelina McMillan-Major and Shmargaret Shmitchell.
26
+
27
+ The aim of the released checkpoints is to boost research on large pre-trained language models for German, especially
28
+ for identifying biases and how to prevent them, as most research is currently done only for English.
29
+
30
+ ---
31
+
32
+ Please use the new GitHub Discussions feature in order to discuss or present further research questions.
33
+ Feel free to use `#gc4lm` on Twitter 🐦.
34
+