Comparing the Random Forest and BERT and GPT3 classifiers for gender predictions from a single tweet.
- Linear SVC: 0.658
- Random Forrest: 0.669
- MLP: 0.697
- GPT3:
- .724 with davinci
- .702 with curie
- BERT: 0.728
Note that GPT3 did not have any training data
You can open the notebook here.