Limitations & Biases: Model Card Update

#3
by Ezi - opened
Files changed (1) hide show
  1. README.md +13 -0
README.md CHANGED
@@ -60,6 +60,19 @@ The supervised training tasks datasets can be downloaded on [Link](https://www.d
60
 
61
  The model could be used to generate lisp inspired DSL code given the human language description tasks.
62
 
 
 
 
 
 
 
 
 
 
 
 
 
 
63
 
64
  ## Training
65
 
 
60
 
61
  The model could be used to generate lisp inspired DSL code given the human language description tasks.
62
 
63
+ ## Risks, Limitations and Biases
64
+
65
+
66
+ As detailed in this model’s [publication](https://arxiv.org/pdf/2104.02443.pdf), this model makes use of the data-set [One Billion Word Language Model Benchmark corpus](https://www.researchgate.net/publication/259239818_One_Billion_Word_Benchmark_for_Measuring_Progress_in_Statistical_Language_Modeling) in order to gather the self-supervised English data samples.
67
+
68
+ Significant research has explored bias and fairness issues with language models (see, e.g., [Sheng et al. (2021)](https://aclanthology.org/2021.acl-long.330.pdf) and [Bender et al. (2021)](https://dl.acm.org/doi/pdf/10.1145/3442188.3445922)).
69
+ As such, it should be noted that language models that are pretrained from text corpus such as the One Billion Word Word Language Model Benchmark corpus have been further explored (e.g by [Ngo, Helen & Araújo et al(2021)](https://www.researchgate.net/publication/355582954_No_News_is_Good_News_A_Critique_of_the_One_Billion_Word_Benchmark) reports that the One Billion Word Word Language Model Benchmark corpus
70
+ > “generate text in the linguistic style of news, without any grounding in the real world. In addition to potential harms from models which are inadvertently optimized for generating fake news.”
71
+
72
+ The aforementioned publication continues to warn that the One Billion Word Word Language Model Benchmark corpus
73
+ > contains sentences which contain words commonly found on blocklists. While these sentences may have plausibly been used in expository contexts within the article, the destructive sentence-level preprocessing and shuffling applied to lm1b [One Billion Word Word Language Model Benchmark corpus] removes all long-range structure from the text and makes it infeasible to track the context and intent of individual examples.
74
+
75
+ [Ngo, Helen & Araújo et al(2021)](https://www.researchgate.net/publication/355582954_No_News_is_Good_News_A_Critique_of_the_One_Billion_Word_Benchmark)
76
 
77
  ## Training
78