Sarcasm detection in news headlines — on cAInvas

Photo by Su for RaDesign on Dribbble

The dataset

Snapshot of the dataset
Class value distribution

Data preprocessing

Code: Text preprocessing functions
Code: Text preprocessing
WordCloud: Not sarcastic
WordCloud: Sarcastic

Train-validation split

Code: Train-val split

Tokenization

Code: Tokenization

The model

Code: Model and training
Test set predictions
Confusion matrix

The metrics

Plot of accuracies
Plot of losses

Prediction

Code: Prediction on test samples
Sample prediction on a random test sample

deepC

deepC compliation

Simplifying AI development for EDGE devices