Model Card for Model ID
The model classifies social media texts as either cyberbullying or non cyberbullying.
Model Details
Model Description
The model classifies social media texts as either cyberbullying or non cyberbullying.
It was built by finetuning a Roberta base transformer and training it on the UC Berkeley measuring-hate-speech dataset.
The model has an accuracy, f1 and recall of 92%.
It was trained using the k-fold cross validation method
The model performs well on explicit hate/harm but misses implicit, coded and slang based harm.
A follow up project is in place/on going to reduce the bias in the model.
-
Developed by: Dianah Naiga
-
Model type: transformer
-
Language(s) (NLP): Python
-
Finetuned from model [optional]: Roberta base
Model Sources [optional]
- Repository: [More Information Needed]
- Paper [optional]: [More Information Needed]
- Demo [optional]: a demo is available on my hugging face space.
Uses
The model is to be used for study and research purposes. It is a very good model for research towards bias reduction in AI models.
The model tends to have racist, homophobic, and sexist tendencies reflecting the bias in the dataset it was trained on.
Direct Use
The model can be used directly as it is already finetuned. You just need to load both the model and the tokeniser.
[More Information Needed]
Downstream Use [optional]
[More Information Needed]
Out-of-Scope Use
The model will not work well for implicit harm, sarcasm, coded language and modern slang.
[More Information Needed]
Bias, Risks, and Limitations
The model was trained on an explicit harm dataset and therefore isnt very good at detecting implicit harm
The model tends to have racist, homophobic, and sexist tendencies reflecting the bias in the dataset it was trained on.
Recommendations
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
Training Details
Training Data
[More Information Needed]
Training Procedure
The model was trained using k-fold cross validation method
Preprocessing [optional]
Tokenisation and space removal was done.
[More Information Needed]
Training Hyperparameters
- Training regime: [More Information Needed]
Speeds, Sizes, Times [optional]
[More Information Needed]
Evaluation
Testing Data, Factors & Metrics
Testing Data
[More Information Needed]
Factors
[More Information Needed]
Metrics
[More Information Needed]
Results
The model has an accuracy of 92%
[More Information Needed]
Summary
Model Examination [optional]
[More Information Needed]
Environmental Impact
Carbon emissions can be estimated using the
Machine Learning Impact calculator presented in
Lacoste et al. (2019).
- Hardware Type: double GPU
- Hours used: 8
- Cloud Provider: Kaggle
- Compute Region: United Kingdom
- Carbon Emitted: [More Information Needed]
Technical Specifications [optional]
Model Architecture and Objective
[More Information Needed]
Compute Infrastructure
[More Information Needed]
Hardware
[More Information Needed]
Software
[More Information Needed]
Citation [optional]
BibTeX:
[More Information Needed]
APA:
[More Information Needed]
Glossary [optional]
[More Information Needed]
More Information [optional]
[More Information Needed]
Model Card Authors [optional]
[More Information Needed]
Model Card Contact
[More Information Needed]
