- Technical area:
- learn transformers
- understand BERT and how to train it through simple transformer
- know how pretrained model work since BERT is one of them
- save pickle checkpoint , load checkpoint and use it for our recommendation system
- tools
- jupyter notebook on vscode
- pandas , numpy
- simple transformer
- skitlearn
- torch
- tensorboard
- soft skills
- communication with teammates
- look at video on youtube
- Achievements
- build classification models : naive bayes, decision tree, logistic regression, SVM
- build advanced recommendation system with BERT
- Next step
- improve the accuracy : cross validation, data augmentation
- try to deploy the model on keroku or GCP