A Classification task with the aim of detect toxic comments from social media using Machine Learning or Deep learning This case study is based on the Kaggle Competition: https://www.kaggle.com/c/jigsaw-unintended-bias-in-toxicity-classification
Detailed Blog: https://medium.com/@vanpariyavishal02/toxicity-of-the-comment-by-jigsaw-1faea0e716b3
Contents:
| S.No | Section | Jupyter Notebook |
|---|---|---|
| 1. | EDA File | jigsaw-eda.ipynb |
| 2. | Baseline Model | Jigsaw-baseline.ipynb |
| 3. | Machine Learning Models | ML_model.ipynb |
| 4. | Model-1 containes 2 LSTM and one attention layer | bi-lstm-attention.ipynb |
| 5. | Model-2 Stacked LSTM with fasttext and w2v | Stacked-lstm-fasttext-glove.ipynb |
| 6. | Model-3 2 LSTM and 2 Attention layers | bi-lstm-bi-attention.ipynb |
| 7. | Feature extraction for BERT | BERT_FE.ipynb |
| 8. | Bert training | BERT.ipynb |
| 9. | End-to-End piepline class | final_function.ipynb |
| 10. | Deployment Code on GCP | Deployment_code.ipynb |