g8a9/ear
Code associated with the paper "Entropy-based Attention Regularization Frees Unintended Bias Mitigation from Lists"
This tool helps researchers and practitioners in natural language processing (NLP) build fairer, less biased models for tasks like hate speech detection. It takes a pre-trained language model and training data, then adds a special regularization technique during training. The outcome is a fine-tuned model that is more robust and less likely to misclassify content due to unintended biases related to specific identity terms. It's designed for those developing or deploying text-based AI.
No commits in the last 6 months.
Use this if you are developing AI models for text analysis, such as hate speech detection, and need to reduce unintended biases that can arise from models overfitting to specific words related to identity.
Not ideal if you are looking for a pre-packaged, ready-to-use API for bias detection without any model training or customization.
Stars
50
Forks
5
Language
Python
License
MIT
Category
Last pushed
May 31, 2022
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/nlp/g8a9/ear"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
xv44586/toolkit4nlp
transformers implement (architecture, task example, serving and more)
luozhouyang/transformers-keras
Transformer-based models implemented in tensorflow 2.x(using keras).
ufal/neuralmonkey
An open-source tool for sequence learning in NLP built on TensorFlow.
graykode/xlnet-Pytorch
Simple XLNet implementation with Pytorch Wrapper
uzaymacar/attention-mechanisms
Implementations for a family of attention mechanisms, suitable for all kinds of natural language...